Thursday, 27 February 2020


It's cold outside, so let's play some video games. Today we're going to have a look at Id Software's Rage, which came out in 2011.

Do you remember Rage? I don't. It passed me by. I probably confused it with FEAR and Prey, which are different games.

What is Rage? It's a first-person shooter with driving sections set in a post-apocalyptic world. Do you remember it now? Probably not, because there are lots of first-person shooters set in a post-apocalyptic world. Some of them don't have driving sections. Rage was the one with megatextures, remember? No?

I bought it recently because I was curious to see what Id Software had got up to after the disappointing Doom 3 (2004) and it was on sale for £2.99. I feel I got my money's worth. At full-price it would have been a bitter disappointment but as a budget game it's great fun. It's also hollow, frustrating, and extremely derivative, and as the reviews often pointed out it feels more like an extended technical demo than a completed game, but when you get past the driving sections and the empty plot the shooting sections are fun.

Nowadays Rage is interesting for a number of reasons. It was the last Id Software game with substantial input from John Carmack, who was part of the team that worked on Doom back in 1993. In fact it was the last Id Software game with substantial input from anyone who had worked on the studio's pioneering titles of the 1990s and early 2000s. Carmack left Id a couple of years after Rage came out, so it was essentially the final product of the old Id. The company still exists, but modern-day Id is really a different company with a couple of executive staff from the old days. The only member of the development team for 2016 Doom that I can name is Mick Gordon, the musician.

I'm not complaining, mind. Rage was the last dreg of inspiration from a team that had run out of ideas, whereas 2016 Doom was a terrific return-to-form. I guess they didn't need John Carmack and John Romero etc after all.

Rage was supposed to show off John Carmack's then-new Id Tech 5 engine. In that respect it was a qualified success. The game has a mixture of indoors and outdoors environments that run at a steady 60fps on the PlayStation 3 and XBox 360, which by 2011 were six years old. It looks very attractive on a high-def television if you're sitting on a couch some distance away, less so if you're using a PC monitor.

Unfortunately I am using a PC monitor. The PC port was widely panned for its narrow fixed field-of-view and lack of graphics tweaking options. It also ran poorly on certain video cards and suffered from a mixture of texture pop-in and slow loading times. A patch cured most of the problems but even in 2019 it feels rough around the edges on the PC.

I don't want to give the impression that I hate Rage. The motion capture and monster AI are very good, the voice acting and writing are functional but far better than they should have been, and as mentioned some of the environments - particularly in the Scorchers DLC mission pack - still hold up today. The core gameplay is solid. The game apparently sold over two million copies, but although publishers Bethesda Softworks tried to make it a "thing" - with a tie-in novel, a comic, a publicity deal with a new NBA player - it didn't catch on and was quickly forgotten by the gaming press and public. A belated sequel came out in September 2019 to widespread indifference.

Bit of backstory. Id Software was founded in the early 1990s. It began as a small indie developer, selling games by mail order, primarily for the PC. Id's early titles had a reputation for technical excellence. They tended to be a few years ahead of the competition - the Commander Keen games of 1990-1991 were fun clones of Super Mario at a time when PC hardware had trouble scrolling the screen around smoothly, Doom (1993) utterly transformed the gaming landscape and single-handedly established the PC as one of the main gaming platforms of the 1990s, Quake (1996) had a fully-3D environment and polgyonal baddies at a time when competing Build Engine games had scaled sprites and fake 3D floors.

The poster on the left is one of the few bits of world-building. At the beginning of the game you get a recorded message from a President Davis, so presumably the asteroid hit during the final stages of an election campaign.

Id's subsequent titles were innovative in their own way, but even at the time I remember feeling that the studio seemed to be running out of ideas. Quake II (1997) was an early killer app for 3D accelerator cards and Quake III Arena (1999) demonstrated that a multiplayer-only title could be a big commercial hit. Id also made money by licensing the Quake game engine to other companies, notable Valve Software, who built Half-Life (1998) on a customised version of the original Quake engine.

Quake III Arena was Id's high water mark. It had strong competition from Epic Games' visually dazzling Unreal (1998) and Unreal Tournament (1999), and although a dozen titles used the Quake III Engine many more used the engine Epic had developed for Unreal. As of 2019 the Unreal Engine is still an industry leader whereas the various iterations of Id Tech are only used by Id. After Quake III the studio seemed to enter a state of semi-hibernation, putting out a new game every five years or so.

Partially this was because the pace of games development slowed in the 2000s. The original Doom was developed by what was even in 1993 a very small team, on a tiny budget; I don't know the exact figure, but major A-list titles of the period had budgets of $1-2 million, and Doom was a shareware game from an indie team, so presumably far less. Ten years later a new wave of high-definition games required dozens of artists and development costs in the $10-15 million range, and ten years after that budgets had ballooned ten times over.

Partially it was because Id found it difficult to retain staff. During the 1990s Id's key programmers were familiar names to computer games fans. Programming wizard John Romero even managed to attract a certain amount of mainstream press attention for his lavish ways, but by the end of the decade most of the people who had worked on Doom and Quake were gone.

Bits of the game are extremely derivative of Half-Life 2.

Designer Tom Hall was dismissed during the early stages of Doom's development, despite coming up with a number of key elements that were used in the final game. Romero was sacked in 1996 during the development of Quake. He went on to set up his own company, which is another story entirely. Designers Sandy Petersen and American McGee brackets his real name close brackets left or were dismissed around the time of Quake II, and programmer Adrian Carmack and designer Tim Willits were let go during the early 2000s, which meant that by the time of Rage John Carmack brackets no relation close brackets was the only one left. For this blog post I have been experimenting with speech recognition, because I talk afterwards the Beatles Boys. Am I more intellectual bye-bye than I type question mark select back back delete talk faster than I type

Id's early games had personality and charm, but as the original staff left the games started to become bland demonstrations of the company's new engine. The original Quake in particular had a monotonous single-player campaign, and although Quake II was more polished in that respect it feels like a dream when I think about it nowadays. I can remember individual bits of it, but I can't remember how the bits fit together point new line

Id's next big thing after Quake III Arena was Doom III, but development took longer than expected, and when the game eventually came out in 2004 it was not the competition-crushing triumph Id envisaged. The pitch-black environments and clever lighting tricks were ambitious, but the gameplay was a monotonous series of jump scares in a series of cramped, nondescript metal rooms. Furthermore it was overshadowed by Far Cry and Half-Life II, which came out in the same year. I wasn't a fan of Far Cry, but its open-world gameplay was refreshingly different and totally unlike Doom III. On a technical level Half-Life 2 was simpler than the other two games, but the visual design was excellent, and in almost every other respect - characterisation, visual design, variety - it made Doom III look simplistic and old-fashioned.

A further and final Quake game, Quake 4, was released without Id's involvement in 2006; I mention it for completeness' sake but I can't think of anything to say about it. I haven't played it. I remember seeing retail boxes in the local computer shop in 2006 and wondering why I was unaware of the fourth game in what had once been the dominant PC first-person shooting franchise.

But what about Rage? What is there to say about Rage? Not a lot. In the near future an asteroid smashes into Earth, but a few people are secreted away in underground bunkers that resemble the vaults in Fallout but are called Arks because Borderlands also has vaults and you can't have three games with vaults, can you? You can't have three games with vaults. That would be silly.

This is all explained in a brief cutscene at the beginning of the game but never really explored. Rage has none of the rich worldbuilding of Fallout or even the stupid-but-evocative details of the Borderlands games. I can't tell if Id wanted you to read the novelisation first, or if they just didn't care. Why was there a novelisation? I realise that there were novelisations of Doom, but why Rage? Tradition? Did they really think that Rage would be a franchise? If so, why didn't they put more effort into the storyline?

Your character is a mute nobody called Nicholas Raine, who is apparently a trained soldier, but the game doesn't tell you any of this. You wake up hundreds of years after the asteroid impact into a world that resembles the Mad Max films. Within the first five minutes of the game you are rescued from mutant attack by a chap called Dan Hagar, who runs a nearby garage. He immediately gives you a pistol and asks you to kill everybody in a nearby bandit base. He doesn't try to ease you into the new world, he just hands you a pistol and tells you to go off and kill a bunch of people. This happens a lot in Rage. No-one asks you about the world you left behind. The game has about as much narrative depth as Quake II, which was fine in the 1990s but feels unsatisfying today.

The ironic thing is that whereas Far Cry had masses of awful writing and terrible acting that went on for ages, the voice acting and writing of Rage is okay, but there's very little of it. The characters are all broad stereotypes, but they are at least unique. It has the typical thing whereby the characters gesticulate wildly as they talk - in exactly the same way that film actors don't - but it's not offensively bad.

The perfunctory worldbuilding was one of the major criticisms of the game back in 2011. Some games feel as if the developers were full of neat ideas, and were prepared to work overtime to implement them, but Rage feels like the kind of thing IBM might have put together if they had been contracted to make a post-apocalyptic shooter with vehicle bits. It meets the specification, and that's all it does.

Rage has a card game minigame. Off the top of my head this is the only time you see your character's face. He's a voiceless, personality-less nothing.

At heart Rage is a linear corridor shooter. You raid a series of bases. To pad things out the game sometimes asks you to raid the same base twice, and in one case you clear a base from back-to-front. The bases have a lot of environmental detail but the background props are almost entirely non-interactive and you can't jump over small objects. Each base has a "track" that you follow. The enemy motion capture and AI is very good, but as with Half-Life 2 you mostly fight the enemy in small rooms and cramped corridors, so the AI rarely has a chance to show off.

You travel from base to base by driving there in your car. Pre-release publicity suggested that the game was an open-world role-playing game, but you can only enter bases when you need to and they're just combat zones. While driving between locations you are attacked by enemy cars, but your missiles generally destroy them in one hit. The driving physics are okay but ultimately the driving aspect feels half-baked. The distances involved aren't long enough to feel satisfying and the car-on-car action boils down to locking on to the enemy cars and firing missiles at them. The game would probably have felt more substantial if there were no cars at all, and you had to travel from base to base on foot - fighting people along the way, or avoiding packs of enemies, who knows - but that would have required more playtesting.

You also have the opportunity to raise some cash by participating in stock car rally races, but there are only a handful of tracks and it's not hard to beat the enemy drivers. The main reward is a form of currency that upgrades your car, which I suppose is a good thing because it means you can finish the driving sequences quicker.

Loosum Hagar is the stereotypical hot farmer's daughter a la Donna Douglas from The Beverly Hillbillies. She was a fan favourite because we live in an ugly world and beauty is rare. She has a much bigger role in the sequel. Incidentally Rage doesn't have a "holster weapon" command, so outside of hub towns you end up jamming your pistol in people's faces when you talk to them. It's another one of those minor little details that makes the game feel older than it is.

There are a handful of hub towns where people give you missions. They're a bit like Megaton or Rivet City from Fallout 3, but there are only a few interactable characters and you don't have dialogue options. Technically you can "accept" or "decline" missions, but if you decline a story mission the game simply stops progressing until you accept. There are no time limits, so the option to decline is pointless. In addition to the main storyline missions there are a few repeatable filler maps and some minigames, but the rewards never justify the time spent playing them.

I don't want to give away the plot, but one of the game's big problems is the lack of a satisfying final battle. Mid-way through the campaign you fight a big mutant monster. It's simple but exciting and the build-up is executed nicely. You see the mutant from a distance first, and then you see its enormous hand as it smashes the building you're walking through, then its feet etc. It's the best scripted sequence in the entire game. After that you have some more bread and butter missions that lead up to the final confrontation with the main villain, except that there isn't a final battle. You do a series of bread and butter missions and then the game ends. You never meet the chief baddy. The Scorchers DLC pack added a second boss battle, but even with that Rage feels unsatisfying and I can understand why it was forgotten so quickly.

The game's saving grace is the combat, but I'm going to digress a bit here. Rage was the first game developed by Id Software primarily for games consoles, in this case the PlayStation 3 and XBox 360. It was later ported to the PC. As such it feels like a console shooting game, in the sense that combat feels more like a tactical exercise than a test of mouse skill. You fight a limited number of baddies at a time, your health regenerates, the level path is linear, and there's a big emphasis on using special weapons when things get tough. You can fling lethal boomerangs and deploy remote-controlled bomb cars, and if you pick up enough background objects you can build replacements. The boomerangs - they're called wingsticks - are one of the best weapons in the game, almost a game-breaker because they kill most baddies with a single fling.

At times Rage feels like an ancestor of the between-arenas sections of 2016 Doom, and in its defence there are some good fights. A simple battle in the Dead City where mutants trap you in a pair of rooms stands out, as does a base assault against a bunch of explosives experts that ends with the base blowing up. Your weapons are generally effective although some of them feel redundant. Near the beginning you pick up a Kalashnikov that is almost immediately replaced by a much more effective shotgun. In terms of realism it's "modern shooter lite", in that you can take slightly more punishment than in a military game - you can even resurrect yourself from the dead! - but you can't afford to stand still for very long otherwise you'll be killed. The baddies tend to be bullet sponges but the aforementioned wingsticks are a godsend.

Id Tech 5's megatexturing technology was designed to eliminate texture tiling...

... without overwhelming the GPU memory of contemporary consoles.

In that respect it works, and the texturing also allowed the level designers to use low-poly models...

...but because this was 2011 and Id Tech 5 was new they went too far at times.

The action picks up towards the end. At the beginning you fight a bland selection of mostly melee monsters who rush you in a straight line, but later on you meet a variety of military goons and metal-clad survivalists who are smart enough to take cover and even retreat when they are outmatched. The shield-generating military soldiers are reminiscent of a similar enemy in Doom and just as irritating.

It has one design decision that irritates me. You can't reload while sprinting. As such the combat often devolves into shooting at a couple of baddies, sprinting away, reloading, then shooting again. It's particularly annoying with the nailgun, which has a very long reloading animation. The game doesn't exploit the vertical dimension as well as Doom, but to its credit the mutant baddies often parkour towards you via ceilings and walls, so combat isn't just a lot of running backwards in figures-of-8 while reloading.

A bit about the technology. Id Tech 5's big thing was megatexturing. Instead of using conventional tiled textures Rage uses enormous, map-sized texture files that are wrapped over the whole level. The engine streams the megatexture from the hard drive, but only streams portions that are visible from the player's point of view, and when the player's viewpoint changes the texture portions that are no longer visible are discarded from memory.

The disadvantage of this approach is that the texture files are huge - Rage is a 23gb installation on the PC, a lot for 2011 - and it would have been impossible to implement megatexturing on the PlayStation 2 or original XBox because they had tiny hard drives. In 2011 however the PlayStation 3 and XBox 360 had 250gb hard drives as standard so storage space wasn't an issue. Distributing the game was another matter; in theory the PlayStation 3 version could have had larger textures, because its Blu-Ray discs could store up to 50gb, but in order to ensure commonality across systems the game's footprint was capped at around 20gb.

Megatexturing has three things going for it. The huge texture files mean that every square metre of the game world can have unique detailing, and Rage is visually very rich. A side benefit of the rich texturing is that the level designers were able to use low-poly environmental models, as in the next couple of screenshots. The detailed texturing disguises the relatively simple models:

From a distance the rocks and pillars look great.

But up close the pillar is just a rectangular block...

And the rock is a relatively simply polygon. Also, the grass in the left of the screenshot is made of flat sprites that always face the player, just like Doom.

Another advantage of megatexturing, although it's a relatively minor one, is that it's easy to mock up static lighting effects by directly painting them onto the megatexture, rather than having to generate a separate set of lit textures. On a pragmatic level megatexturing also simplifies file management; the artists no longer have to deal with hundreds of individual texture files. A side-effect of this is that megatexturing in theory should eliminate texture errors, because they would be obvious from looking at an overview of the megatexture.

Unfortunately in Rage the megatexturing is only partly successful. It ran smoothly on contemporary consoles, and some of the maps are very good-looking; the Subway Town hub that you visit late in the game would have been a fantastic base in a fully-fledged role-playing game. However the game is plagued by texture pop-in. My PC has eight times as much texture memory as a PlayStation 3, but when I turn the camera left and right objects at the sides of the screen draw themselves into existence, and when I turn a corner the wall in front of me shimmers for a moment as the texture is decompressed. It only happens for a split-second, but it's visually jarring.

The second problem is the low texture resolution. In order to keep the file size within reasonable limits Rage's megatextures were heavily compressed. At a distance everything looks fine, but up-close the environments are grainy and low-detail. On top of this, in order to maintain sixty frames a second a lot of the world objects are very crude, and the game's environments are almost entirely static. The boxes, chairs, and miscellaneous clutter that litter the floor are all built into the maps. You can't knock them over. Rage is perhaps the first post-1990 3D game I have played where you cannot open the toilet stalls. The game also has a very simple static lighting model - each room has a colour filter, but that seems to be about it - albeit that this isn't a problem with megatexturing per se.

Furthermore the megatextures implemented in Rage rarely have animations, so whereas other games are full of blinking lights and animated computer screens the background objects in Rage don't do anything. In the following screenshots the computer screens are low-definition static textures and the wall machinery isn't animated, so it looks very poor:

Incidentally that location has a dialogue sequence, but it's coded in the expectation that the player would stand in front of these two characters:

If the player stands to one side, as I have done, the player model clips into one of the characters when the dialogue is over:

A bit more testing would have picked up that issue. Anything else? The world is covered in what looks like film grain. I think it's a noisy texture designed to make the world look gritty, but it doesn't work.

On a purely technical level the visuals of Half-Life 2 were functional even in 2004, but the aesthetic has aged well.

Ultimately Rage looks odd. It has the environmental texturing and detailing of a modern game combined with the low polygon models and monolithic, non-physics-enabled environments of something from the 1990s. And yet the character animation is still very good, so it confuses my brain. When you kill the baddies they don't just drop straight to the ground, they tumble forwards or recoil backwards depending on how they were moving before they died. That element of the game works very well.

On the whole megatexturing as implemented in Rage simultaneously feels over-the-top (most of the indoors sections could have been tiled without losing anything and the bland desert environments look very similar) and underdeveloped (couldn't they have layered a few animated textures on top of the megatextures?). As mentioned up the page Rage feels like a demonstration of Id Tech 5 that was hobbled by the limited memory of contemporary consoles and a lack of development time.

The original Quake was infamous for its monotonous colour palette - brown and purple. Rage adds cyan and orange.

In another world modders might have fixed Rage's problems, but megatexturing makes it difficult for fans to create home-made levels. Without access to a common pool of texture tiles hobbyists have to generate their own megatextures; the texture resources for Rage were apparently rendered on a server farm. Id released a suite of development tools two years after Rage came out, by which time no-one cared about the game any more. There doesn't appear to be a Rage modding community.

One last thing before I stop writing. In 2011 reviewers pointed out that Rage felt like an anonymous imitation of Borderlands and Fallout 3, which are also set in run down post-industrial environments. Rage is also extremely derivative of Half-Life 2. One level is a straightforward clone of Half-Life 2's Nova Prospekt prison level, complete with contrasting high-tech architecture and armoured soldiers who deliver radio patter as they flank the player. The armoured soldiers even look like Combine Overwatch soldiers, but painted black and red instead of white. Playing the latter levels just reminded me of how much character Half-Life 2 had.

Subway Town is the game's other major hub. Even today it's still visually impressive. The gameplay and missions however are just a copy-and-paste from Wellspring; you suck up to the mayor, upgrade your car, and then the baddies take over.

The game has a soundtrack, but it's not very good and pales in comparison to Mick Gordon's work on Doom. The action themes are short and loop around without much variation; the driving sequences have a short guitar theme that quickly gets boring.

Can I think of anything else to say about Rage? No, I cannot. It's an anachronism that combines the gameplay of a good early-2000s shooter with the environments and character models of something much more modern, plus the environmental detail of something from the late 1990s. It takes place in a world where physics-based gameplay never happened, where games do not have storylines.

An exceptionally rare physics-enabled object

Also, why is it called Rage? I have no idea. The main character isn't angry. He never says anything. He's infused with nanomachines, but they don't make him angry. The baddies are no more angry than any other video game baddies. Did the game originally have a different storyline? Did they just pick a snappy name out of a hat?

As before, if you know the answers to these questions, or any other questions raised throughout this blog post, please send your answers on a postcard to Iain Duncan Smith care of the Home Office. Just write "yes" or "no" on the postcard. Don't include a return address. It's probably a good idea to handle the postcard with kitchen gloves, even if this means wearing kitchen gloves when you buy the postcard. No-one will mind because it's 2019 and alternative lifestyles are perfectly normal.

Saturday, 15 February 2020

Canon EOS-1Ds: Mother is the Breadmaker, Father the Yeast

Let's have a look at the Canon EOS 1Ds. The original 1Ds, from 2002. Back then it was a heck of a thing. It was built to deliver blunt force trauma; hurting bombs. Heavy-duty, cast-iron piledriving punches. It had an eleven megapixel sensor at a time when people were still impressed with six. A full-frame sensor at a time when people were still amazed by the smooth tonality and narrow depth of field of APS-C, and it was all wrapped up in the best camera body Canon could make.

In 2020 age has withered it, and custom staled its infinite variety, but on the whole it hasn't dated all that badly. The autofocus and metering are still impressive, and on a physical level - as an object - the body looks fantastic. It falls apart under artificial light, and its performance is surprisingly mediocre at higher ISOs, but in daylight it produces noise-free images with subtle colours. Operationally the body is instantly responsive but the I/O bus is very slow. Let's not talk about the LCD monitor.

In the early 2000s Canon seemed to be much less enthusiastic about the digital SLR market than Nikon. The Nikon D1 was a big success with photojournalists, and on the whole Nikon's range of digital products - its swivel-bodied CoolPix cameras, the Coolscan film scanners, the early digital-friendly 17-35mm f/2.8 - felt like part of a thought-out system. Canon had a strong presence in the consumer market but its professional digital portfolio felt haphazard. In that context the 1Ds was a powerful statement of intent; its combination of resolution and large sensor were a generation ahead of the competition.

Judging by its factory code my 1Ds was built in December 2002, in Canon's factory in Oita, Japan. The shutter is rated for around 150,000 shots; mine has taken 27,000; the lubricants will probably gum up long before the shutter breaks. The capacitors will rupture. The rubber will perish, the LCD backlight will burn out. By that time no-one will service it - as far as I can tell no-one services it in 2020 - and humanity will have mutated into a form that no longer uses visible light to detect predators. It will have a second life as a boat anchor.

Canon's in-house competition was the Canon 10D, which had a six megapixel, APS-C sensor. It was launched slightly before the mid-2000s digital SLR boom, and for a few months in 2003 Canon only sold three digital SLRs - the 1D, the 1Ds, and the 10D. To confuse buyers they all had different sensor formats (APS-H, APS-C, and 35mm).

By the year 2002 several manufacturers had experimented with full-frame sensors, but none of the first generation of full-frame digital SLRs had been unequivocally successful. The 1Ds bucked the trend; it was mostly quirk-free, and despite a price of around £7,000 / $8,000 it apparently sold well.

I don't have access to Canon's sales figures, but the unit reviewed by Imaging Resource in 2002 was serial number 101003. Mine is 117893, so assuming an unbroken run of serial numbers Canon must have sold at least 16,890. It's unlikely that mine was the last ever made, so I guess there were 30,000 or so, a lot for something aimed at such a small market. Very few people on the internet have photographed the bottom of their original 1Ds, so I can't make a more accurate guess as to how many were made.

The 1Ds' body and autofocus system were shared with the EOS 1D, which had an APS-H sensor, and were apparently derived from the EOS 1V, which was the last of Canon's professional 35mm film cameras, so presumably the most expensive part of the 1Ds' development process was the sensor, which was a CMOS unit of Canon's own devising.

A 1Gb IBM Microdrive - $400 or so in 2002 - stores 87 RAW images.

Writing about the original 1Ds is difficult. As far as I can tell the camera sold almost exclusively to press agencies and moderately well-heeled wedding, portrait, and advertising photographers, who were for the most part focused on their business than writing about cameras on the internet. Their work was not intended to last beyond the moment. The leading fashion and editorial photographers of the day used whatever the heck they wanted, ranging from autofocus point-and-shoots to large format view cameras, and medium format photographers were not impressed with eleven megapixels.

Furthermore an awful lot of photographers still shot film in 2002 - not because it was fashionably retro, but because it was normal - and so my hunch is that the 1Ds was embraced by a mixture of forward-thinking people who were not sentimental about the past and middle-class computer nerds. Hard-nosed hard-charging pioneers, who at this moment in time have probably switched to drone photography, or have abandoned traditional cameras in favour of high-resolution mobile phones.

Bear in mind that in order to use the 1Ds effectively in 2002 a photographer needed a bunch of memory cards, which were not cheap, plus a decent computer - also not cheap - and a bag full of batteries and lenses, plus some flash units or studio strobes. The toppermost of the poppermost would have had a second 1Ds body plus an Apple laptop so that they could transmit images in the field, again not cheap. The 1Ds was less expensive than a digital medium format back and a lot more flexible, but still very expensive as a system.

By the time the 1Ds entered the second-hand market Canon sold the smaller, more competent 5D, which drew attention away from it. Most of the leading photo websites existed in 2002, but beyond first-look reviews there isn't much on the internet about the original 1Ds. I have the impression that most of the surviving bodies are now objects d'art.

Off to Lidice, a village in the Czech Republic.

One of many villages obliterated by German forces during the Second World War.

They destroyed it in June 1942 in retaliation for the assassination of Reinhard Heydrich. The Germans appear to have selected Lidice on a whim; the evidence that it harboured members of the resistance was flimsy.

They shot all the men and sent the women to die in the concentration camps. The children vanished into the night and fog.

Lidice was bulldozed. After the war it was rebuilt a few hundred yards from the original location. A handful of the original residents returned. Heydrich was buried in Berlin, in a grave marked with an elaborate SS tombstone, but the Allied forces removed almost all trace of it at the end of the war. Lidice lives on.

The 1Ds wasn't the first full-frame digital SLR, or even the second. Or the third. Or the fourth, although it was the first unequivocal success. The first generation of full-frame cameras were all based around a six megapixel sensor made by Philips, the FTF3020, which was introduced in late 1998. It was built from a matrix of 12x12mm one-megapixel sensors ganged together. Apparently Philips had terrible trouble mass-producing the FTF3020 to a sufficiently high quality.

A glimpse of the 1Ds' big sensor. In common with most digital SLRs from before 2005 or so the sensor cleaning mode simply flips up the mirror - you have to remove dust from the sensor yourself.

The Philips sensor appeared in the Contax N Digital of 2002, which was released in limited quantities and has left very little trace on the internet - this review is probably fantastic if you read Thai - and the Pentax MZ-D, which was developed into a functional prototype but cancelled before release. The N Digital had a base ISO of 25 but had shadow noise even at that low level.

The sensor also appeared in a number of 6x7 and 645-format medium format digital backs from Leaf, Sinar, Phase One among others, where it would have had a crop factor. It would be fascinating to try one of these cameras out in 2020, but getting the right combination of sensor back, camera body, tethering cables, batteries, and period-correct software would be extremely difficult.

By the time the 1Ds came out the first generation of full-frame cameras had died off. In 2002 Philips sold its CCD business to DALSA, who concentrated on the scientific market. In 2002 the 1Ds' most direct competitor was the fourteen megapixel, full-frame Kodak DCS Pro 14n, which was built on a Nikon F80 film body. I have one! Here's what it looks like, on the left:

This picture doesn't really get across how fat it is from front to back. It's an ugly fat lump. A big ugly fat lump. A camera that, if you saw it while eating, you would be unable to finish your meal. A camera that, if you saw it in the street, you would point it out to your friends so that they could join you in mocking it. A camera that dedicated its life to making money by lying with every breath it could muster in order to keep murderers and rapists on the streets. It tried to play husband. It tried to taste the life of a simple man. It didn't work out.

It has just dawned on me that Se7en is twenty-five years old. A quarter of a century. Yikes. I've always had a soft spot for the Kodak DCS cameras. In the 1990s Kodak had to work everything out for themselves, and they mostly got it right, but the 14n is hard to love. In its favour it has fourteen megapixels versus the 1Ds' eleven, and it doesn't have an antialiasing filter, so the images look crisp at 100%. It has two memory card slots - Compact Flash and SD - and it will write any combination of RAW and JPG images to either card, whereas the 1Ds has a single CF card slot and only writes RAW or RAW+JPG. There's no option to just output JPGs.

Against the 14n anything higher than ISO 160 has unpleasant shadow noise, the F80 camera chassis is unimpressive and feels flimsy, the sensor can't cope with long exposures beyond two seconds, images tend to have a peculiar green/purple colour cast, and overall the colours feel flat and purple, which is odd for a camera engineered by Kodak. At launch the SD card slot was disabled because Kodak hadn't finished writing the camera's firmware. The slow card write times, low base ISO, and low flash sync speed (only 1/125th) made it awkward for wedding photography.

At around £4,000 the 14n was much cheaper than the 1Ds, and if you had a bunch of Nikon lenses it was the only full-frame Nikon-mount camera until the D3 in 2007. Kodak tried to fix some of the camera's problems with the Pro SLR/n and SLR/c a year later, but it wasn't enough, and the company left the professional digital photography market in 2005. As of 2020 it's difficult to use the 14n extensively because the batteries are long out of production. The 1Ds has a similar problem but third-party batteries are still available, so it will continue working for a few years yet.

The 1Ds used the same batteries as the 1D. They're Nickel-Metal-Hydride (NiMH), an older technology than modern Lithium batteries, in turn more advanced than the Nickel-Cadium (NiCad) batteries used by the earlier Kodak DCS cameras. They weigh a lot and last for hundreds of shots. The charger is a big block that connects to a pair of batteries with tendrils.

I have the impression that back in 2002 most reviewers assumed that full-frame was an inevitability, and that the 1Ds would be followed by a flood of similar cameras, but in practice Canon had the field to itself for several years. The next full-frame digital SLRs were Canon's own sixteen megapixel 1Ds MkII (2004) and the compact, twelve megapixel 5D (2005). In the meantime Nikon stuck with APS-C, and for a few years the company seemed to lag behind, but they had a strong comeback with the full-frame D3 (2007), which prioritised low noise over high resolution.

I mention all this because the 1Ds wasn't just a camera. It was also a potent expression of Canon's technical dominance, and also a great status symbol for Canon and the photographers who had enough coin to own one or use one for their job. Nowadays digital SLRs that don't shoot video are old-fashioned, but seventeen years ago the 1Ds would have opened doors. It would have won arguments and one-upped the competition.

Does the original 1Ds make any sense in 2020? Not really, but it's more obsolescent than obsolete. The clicky-clicky camera side of things is still very good. The 45-point autofocus system is fast and precise. It has a bunch of autofocus points that become super-accurate at f/2.8 and wider and I feel slightly ashamed that I will probably never tax the 1Ds autofocus system. For most of the images in this post I used a Canon 100mm f/2, and I didn't have any problems focusing on things:

The red channel seems to blow out green.

Focus tracking? I have no idea. I didn't encounter any race cars. The autofocus hardware was used in all of Canon's professional-level cameras from the late-1990s up until the 1Ds MkIII of 2007, so I have to assume it was good enough for the world's sports photographers. Flash sync speed is 1/250th, which is good but not exceptional, but the camera is compatible with E-TTL Mk1, which supports a strobed high-speed-sync mode on certain flash units.

The first two things that struck me about the 1Ds were the hairtrigger shutter button and the lighting-fast mirror blackout, which at 87ms is literally faster than a blink. The mirror seems to go FLICK rather than KER-CHUNK. In fact it doesn't go FLICK, it goes FLK. Got that?

Furthermore the body has a bunch of weather seals. Sadly the I/O port cover on my example has been pulled off, a popular modification at the time; I suppose I could cover it with gaffer tape. The body itself feels like a single block of metal with textured plastic on the handgrips. The only weak point is the battery, which - as with the Nikon D1 - forms the bottom-left part of the case. You have to twist a knob and press a button to remove it. The batteries have rubber seals to maintain the camera's weather proofing, but the outside cap is plastic. I have seen some used 1Dses with bashed-in prism housings, so it's not indestructible, but I imagine it would survive a drop to a concrete floor without smashing apart.

You have to move your hands an awful lot to operate the camera. The menu system is reminiscent of the Kodak DCS cameras, in that you hold MENU while turning the rear dial to move left and right between menu pages, and then SELECT and the dial to move up and down through the menu options. The top plate has three weatherproof buttons that operate in combination with the rear and (mostly) forefinger dial:

From top to bottom you can change the shooting mode (AV, TV, etc), the autofocus mode (tracking or one-shot), the metering mode and flash exposure compensation, and then by holding two buttons you can change the bracketing interval (up to three stops), the ISO, and whether the camera fires in single-shot or burst mode. I changed ISO most of all, because the camera doesn't have an auto-ISO mode.

The ISO ranges from 100 to 1250, an odd top figure that was shared with the APS-C EOS D60. I'm not sure why it didn't go up to 1600. There's also a menu option to enable ISO 50, but I think it's a software creation that simply overexposes the image by a stop and then de-exposes it during development. The metering tends to err towards underexposure, or at least when confronted with an outdoors scene it tries to retain highlight detail in the clouds, which is fair enough. My hunch is that the exposure system was modelled on slide film, where you have to expose for the highlights. Images from the 1Ds benefit from having the shadows boosted, which increases grain, but at lower ISOs it's not all that noticeable.

This is ISO 1250, with the shadows boosted by about a stop. The noise isn't quite random, but sized down like this it's not offensive. The 1Ds doesn't cope well with fluorescent light - if you correct for the orange cast the image turns purple instead.

I took this quickly, while crossing the road, but left to its own devices the 1Ds did a good job of keeping detail in the clouds. I've boosted the shadows with Photoshop.

The big selling point of the 1Ds was its huge file size, almost twice that of the contemporary Canon D60, 10D, and Nikon D100. By modern standards eleven megapixels is small for a stills camera - it's only slightly larger than 4K - but the 1Ds produces sharp, detailed, smooth images that are still pleasing to the eye. It has a relatively weak antialising filter that nonetheless doesn't generate very much moire. Here are a couple of examples of the detail it can produce:

Again, the metering system retained detail in the clouds, and I had to brighten the lower part of the image with Photoshop. I left my grad filters at home. This was shot on the observation deck of Prague's TV tower, looking towards the castle.

As a consequence there's a bit of grain at 100%, but it's not objectionable. There's a tiny, almost unnoticeable touch of false colour moire in some of the fine details - the windows on top of the tower - but you have to look very closely to see it.

This is a reverse-angle shot taken from the vicinity of Prague's castle. You can see the TV tower on the horizon, on the left.

At the bottom of the image, to the left, you can see the yellow building in the photograph above - the two towers are just poking over all the other buildings. Again, I've brightened the image in Photoshop to bring out detail in the shadows, so there's more grain than there would be if I used manual metering and a grad filter.

The 1Ds shoots at three frames a second and has a ten-frame buffer. It does however take a while to write images to the card - that's one thing that separates it from the modern age. It wasn't aimed at sports photographers but I imagine that in its day professional photojournalists could work around the slow buffer. On a physical level the 1Ds is heavy, about 1.6kg, but it balances well with long lenses.

As mentioned earlier the camera side of the 1Ds - autofocus, exposure, handling etc - is still very good today, although I miss auto ISO. The sensor however has noticeable shadow noise and only retains about a stop of highlight information, so even with graduated filters and careful exposure there will inevitably be instances where you have to bracket.

In its favour it copes well with long-duration exposures. The 1Ds' contemporaries from Nikon and Kodak weren't very good in that respect - my 14n spits out black frames if the exposure is longer than two seconds, and from my experience of using a Nikon D1x the images were usable but plagued with hot pixels if the exposure was longer than ten seconds or so.

The 1Ds uses a dark frame system, whereby e.g. a thirty-second exposure is followed by a second thirty-second exposure with the shutter closed, which is then subtracted from the original. The following image is a thirty-second exposure taken with the lens cap on:

If I boost the contrast there's some low-level banding, which would be invisible ordinarily, unless perhaps you were taking pictures of distant stars:

My 5D MkII is no longer state of the art, but it has a more advanced sensor than the 1Ds. If I try the above trick - a thirty-second exposure with the lens cap on, boosted with Photoshop's "auto contrast" - I get the following result:

It looks more purple because the 5D MkII has twice as many pixels. It's not completely random, but random enough that in real images it would be unnoticeable. I was just curious to see what the result looked like.

The 1Ds has a large, 100% viewfinder. In the following image I arranged the four pipes along the bottom of the viewfinder frame when I took the picture, and that is where they were when I opened the image in Photoshop.

Prague's main train station has a little vintage station on top of it. Until recently there was a cafe and shops, but as of early 2020 it's deserted.

The 1Ds was launched on the cusp of The Modern Age, so for the most part it doesn't have the odd quirks that make so many early digital cameras charming. It has a conventional Bayer-pattern colour sensor, with non-removable antialiasing and infrared blocking filters. It appears to have no problem with large memory cards, at least up to 32gb. A 32gb card will store 2570 RAW files.

It does however have some oddities. In common with the first generation of professional digital SLRs it doesn't have USB, it has Firewire:

It only has three ports. Firewire, Canon's proprietary remote socket, and PC flash sync.

Why Firewire? In 2002 USB 2.0 was relatively new and photographers all had PowerBook G3s. That's my theory. The Firewire port is awkward nowadays. If you just want to transfer images from the 1Ds it's quicker to take out the card and use a card reader, but the 1Ds has a number of personalisation options that can only be set by plugging the camera into a computer and running Canon's EOS Utility software.

The interface is reminiscent of the Kodak DCS models, but with a splash of colour. It was completely overhauled for the 1Ds MkII.

The camera has the standard array of custom functions for autofocus and button assignments etc, but it also has a range of personal functions that go into more detail. With the personal functions you can turn off most of the exposure modes, so that the camera only shoots in program autoexposure (for example), or you can lock it to only fire at certain shutter speeds and apertures.

I'm not sure why; perhaps it was aimed at photo news agencies who wanted to make the cameras idiot-proof, or studio photographers who expected to mount the 1Ds on a sturdy tripod and shoot thousands of images at f/11, 1/250th with studio flash, and they were worried that an assistant might change some of the settings. Who knows.

Very few PCs ever had Firewire, and Apple dropped it in 2012, but luckily I have a vintage Power Macintosh G5 and an old Canon EOS Utility disc, so I connected my 1Ds to it:

Firewire 400 was apparently on a par with USB 2.0, speed-wise - as a PC person it passed me by - but as mentioned in the text it's quicker to grab images from the card than directly from the 1Ds.

Some of the options are very useful. In this case you can set the camera to continuously churn through HDR bracketing. Of note you still have to turn the functions on with the camera's interface - EOS Utility merely makes them selectable. They're greyed out otherwise.

Quiet operation is disappointing. It makes the quality of the shutter noise slightly different, more spread out, but not really any quieter.

There's also a personal function that enables image zoom. I'm not sure why that isn't turned on by default. Anything else? EOS Utility can embed your name in the camera. Mine was blank, so I have no idea who used to own it.

On the front of the camera, just above the EOS-1 logo, there's a translucent white oval. This has an external white balance sensor behind it. The Kodak DCS models also had external white balance, as did a couple of the early Olympus Four Thirds cameras. I can't say that the 1Ds' white balance is noticeably better than any other camera I have used. It was dropped from the 1Ds MkII and hasn't reappeared, so I assume it was more trouble than it was worth.

The 1Ds saves its RAW files with a .TIF extension. They're around 9-10mb each. Kodak also used .TIF extensions so I wonder if Canon simply assumed it was the standard. In practice they're just standard Canon RAW files - Photoshop processes them just fine - but I imagine that the extension might confuse some operating systems. Thankfully, and unlike the D30 and other cameras, the 1Ds doesn't generate tiny little thumbnail files and it doesn't split its files into 100-file folders.

In common with other professional-level cameras the 1Ds has a physical shutter that closes off the viewfinder. You have to pull down a little lever just to the right of the viewfinder. It's intended to stop light entering the camera during long exposures, especially during daytime.

Anything else? The 1Ds takes a fraction of a second to boot up. It supports a range of different focusing screens, with the Ec-C III Original Laser Matte as the standard. It doesn't have Live View and it can't shoot movies. Those things were several years away in 2002. As mentioned earlier you have to clean the sensor manually.

In-camera HDR? Electronic spirit level? Wi-fi? No, no, and no. You can upload different tone curves to the camera but there are no JPG processing options otherwise. Some of the focusing screens are matted for 5x4 and square format but you still have to crop the images manually. As far as I can tell the only unusual accessory was Canon's Data Verification Kit, which could be used in conjunction with one of the custom functions to add a digital watermark to the camera's JPG images; the watermark was supposed to prove that the image hadn't been altered.

The 1Ds also has a voice recording function, which was something else that appeared on the Kodak DCS cameras. You can associate one or more thirty-second .WAV snippets with each image. This makes sense if you're a photojournalist in the field, and you need to take notes; it's not very useful otherwise.

The camera shipped with the DCK-E1 battery coupler, which had a dummy battery that could run the camera from mains power. As far as I can tell Quantum et al didn't sell an external battery pack for the 1Ds; the batteries are however "dumb", so there would have been no obvious technical obstacle to doing so.

What happened? To the 1Ds? What happened to the 1Ds? It was replaced in late 2004 by the 1Ds MkII, which had a sixteen megapixel sensor but was otherwise very similar. A year later Canon launched the relatively compact twelve-megapixel 5D, which had a full-frame sensor in a much smaller body than the 1Ds. It sold for less than half the price of the MkII, £2,500 vs £6,000, and stole away some of the 1Ds' market share. For a lot of photographers weather sealing and high-sensitivity multi-point autofocus were superfluous and the 5D was good enough.

The 1Ds MkII was in turn replaced by the twenty-one megapixel 1Ds MkIII in 2007, which introduced live view and automatic sensor cleaning, and had a 5fps burst mode. Contemporary reviews were positive although I have a sense that the market wanted something more than just an extra five megapixels.

Ultimately the MkIII was the last of the 1Ds cameras. The market changed abruptly at the end of the decade, for a number of reasons. Digital videography took off, and in that context having a high resolution sensor wasn't very important; at the same time computing power increased to a point where Canon could merge the high-speed 1D press cameras and high-resolution 1Ds studio cameras into a single machine, and as a consequence the line continued with the 1D X in 2011, which had an eighteen megapixel full-frame sensor that could shoot at fourteen frames a second. Furthermore the success of the Nikon D3 demonstrated that a lot of photographers were perfectly comfortable with a lower resolution if it meant they could shoot at ISO 6400.

A year after the 1D X Canon released the similar, video-optimised 1D C, by which time the stills-only 1Ds felt very quaint. Its modern conceptual equivalent is the fifty megapixel 5Ds - crushing resolution superiority with a full-frame sensor - although on a detail level they're very different cameras.

To an extent the high-end digital SLR market has also been struck by the rise of smartphones; what is the point of high-resolution portrait photography in a world where fame and fortune can be made from low-resolution, crudely-filtered Instagram pictures?

This was shot with my mobile phone. The highlights are blown out, but the image is otherwise bright and vivid. I wasn't stopped by security guards and after taking the shot I could immediately upload it to Instagram.

I'm digressing here. Does the original 1Ds make any sense in 2020? Not really. It's not bad, it's just far too big and heavy for what it does. Back in 2002 photographers were prepared to put up with the size and weight because they wanted eleven megapixels, but in 2020 that doesn't apply any more. In its favour it's still very robust, and would make a decent events or possibly even sporting camera if you picked the moment carefully.

On a pragmatic level the body still looks impressive, so if you're trying to pose as a professional generate an air of professionalism you can claim you bought it in 2002 and still use it for sentimental reasons.

If you're too young for that, pretend you bought it from the man who did all the on-set publicity photographs for Harry Potter and the Prisoner of Azkaban. Say that he went to Scotland and took loads of photos of Alfonso CuarĂ³n with it. Make something up. No-one will know otherwise. Remember that Derek Trotter was a moron whose ill-thought-out schemes continually failed, but he was the hero and people liked him. Boycie was much better off and did no-one any harm, but he didn't have charisma so he was the villain. Charisma will get you places.

On the negative side, for concert photography the 1Ds' shadow noise and poor performance under fluorescent light are awkward. For hiking and landscape photography it's weather-sealed but very heavy. The batteries are also very heavy, and NiMH cells perform poorly in cold weather. The charger is also very large. You could alternatively use a much smaller camera with an underwater housing, or a plastic bag, or some kind of rubber weather shield. Bear in mind that top mountaineers of yore carried Olympus OM 35mm SLRs and just wiped off the snow.

Alternatives? If you want to shoot action sports on a budget the APS-C Canon 7D MkI sells for a similar price on the used market, and has a comparable autofocus system, more resolution, movie mode, and weather seals. If you want high resolution a used 5D MkII sells for slightly more, but has twenty-one megapixels and movie mode albeit that the autofocus system is nothing special. If you want to experiment with full-frame an original Canon 5D is a lot easier to carry around.

In the mid-2000s there was a debate as to whether a used 1Ds was a better buy than a new 5D, and in the late 2000s there was another debate as to whether a well-used 1Ds was better than a lightly-used 5D. Resolution-wise the 1Ds apparently has a lighter anti-aliasing filter, but in my experience the 5D's sensor was noticeable newer, better under poor lighting, decent at ISO 1600, so I would opt for that instead.

That's almost all I have to say about the 1Ds. If you want one nowadays be sure to get the NC-E2 charger as well, because there aren't any third-party alternatives. Canon's website still has the manual, but irritatingly it doesn't host Digital Photo Professional or EOS Utility, only updates; you need to either dig out an old CD or download a copy unofficially. The 1Ds should be fully compatible with modern EOS flash units and lenses, although it will only run flash units in E-TTL I mode (E-TTL II, the current system, was a minor upgrade).

At least in the UK Canon's professional servicing network assigns points to different pieces of Canon gear, depending on a formula known only unto Canon; the more points you have, the better service you receive. In 2002 the 1Ds was probably worth a million points, and would have been enough to have Canon send either Gillian Anderson or David Duchovny or both to your doorstep with replacement equipment on a plush golden pillow plus some handcuffs and champagne.

In 2020 however the original 1Ds is worth 80 points. All three of the 1Ds models are worth 80 points. In comparison the 5Ds is worth 290 points and the 200mm f/2 IS is worth 500. The 1Ds still has a little bit of glory, a bit of faded glory, but only a little bit.