I wrote this post quite a while ago, in 2008, and camera tech has improved dramatically. Now I use the 5d Mark III, which is tons better than the Mark II. You can see the inside of my gear bag here. I shoot primarily mountain weddings in Vail, Beaver Creek and Aspen, and I like to do lots of couples portraits outside, at night. When you’re shooting in big cities, it’s not that problematic to have poor autofocus, since there’s a lot more ambient light than there is under the moonlight at 11,000 feet on the side of a mountain. Anyway, the 5d Mark III offers incredible autofocus (yes, it can focus under a full moon) and far superior ISO capabilities.
ORIGINAL POST
Part I
let me issue a disclaimer: i’m a long-time canon user, but for no particular reason other than i started with canon at a young age. i have no allegiance one way or another to nikon or canon, so this 5d mark ii review isn’t going to be tainted by that.
i had been super excited about the arrival or the 5d mark ii and its much-heralded 2 to 3 stop noise advantage over the original 5d, which i’ve been using professionally since its release 3 years ago (and, by the way, can be had for a song on amazon right now). so naturally the first thing i did last night when it came as to do some high-iso comparisons with my workhorse mark i bodies. and guess what? i was sorely disappointed. this thing does NOT live up to the hype, imho. i will likely be branded a heretic by many, but so what. the 5d mark ii is a marginal improvement *at best* over the original 5d when it comes to high-iso performance. i might go so far as to say its massive file sizes will only serve to clog my workflow and create problems, rather than solve them. but i will hold off on clinging to that statement until i shoot more with this new body.
i shoot RAW only, sort, and then edit in lightroom. i make wedding albums with spreads up to 30″ wide, and i make canvasses up to 40×60. i have done this with spectacular results in all lighting conditions with my original 5d. so the question is: do i need more resolution? doubtful. i was really looking for better AF accuracy and low light iso capabilities.
for this review i decided to upres the 5d files to the same dimensions as the 5d mark II files. this is about noise on the second generation body and if i were to downres the 5d mark ii files, i’d be crunching some of the noise out. and besides, when i do make albums, i often upres the mark i files anyway to make spreads. so i think it’s fair to present the images this way as it’s a close approximation of real world use.
so without any further ado, i present to you my thermostat. this was the first test i did last night. the room was lit with natural window light and i shot at av with a canon 50mm at f/2, raw. one is the 5d at 3200 and the other is the 5d mark ii at 6400. these are center crops from full res images, by the way. they’re about 1600 pixels wide out of a 4200 pixel wide 5d mark i file and a 5600 pixel wide 5d mark ii file. both were processed in lightroom with out any fiddling other than luminance and color noise reduction set at 20. go ahead, open up the file and take a guess which one is which.
i happen to think the bottom one looks better, though it’s ever so slightly less sharp. the top image has some banding in it. tight film-like grain, but an awful lot of it. i choose the lower one.
the top image is the 5d mark ii at 6400. i was expecting its images at 6400 to rival the 5d at iso 1600 or even iso 1000. i didn’t expect them to look worse than the mark i at 3200. i think iso for iso they’re about the same, with the 5d ii having perhaps a 1/3 to 1/2 stop advantage.
sure, it’s 21 mp instead of 13 mp, but we will address that later.
anyway, this morning i decided to do more comparisons. so i present the inside of my closet. iso 800, 1600 and 3200. the fourth image below is iso 6400, H1 and H2 on the mark ii. i shot the following with the canon 50mm f/1.4 at f/8. it’s not an L lens, but stopped down that far it might as well be.
oh, and let me be as explicit as possible: these images are the same pixel dimensions because the mark i images are upscaled to the same size as the mark ii images.
draw whatever conclusions you wish. mine are:
1) the 5d mark II does not live up to its hype
2) at any given ISO the mark ii doesn’t look that much better to me on the screen
3) H1 is silly
4) H2 is stupid
other thoughts…
first, i have been reading reports that adobe really screwed up with its 5d mark ii raw module in lightroom and ps. that may be true. it also may be true that canon’s dpp renders a far better jpeg. but switching workflow to the outdated, old school dpp is not feasible, especially when it means losing the localized editing abilities and other huge benefits of lightroom. so if it’s adobe’s fault, i say to adobe: rewrite your raw converter.
second, i have a feeling that a lot of the reviews i’ve seen were done with jpegs and in-camera noise reduction. that’s nice and all, but it’s not how i or many of my colleagues do our work. i shoot in raw and edit in LR like thousands of other people. i do not use in camera noise reduction on jpegs. so that might be why i have been seeing some amazing looking high-iso images elsewhere. if that’s the truth, and if dpp really does a better job compared to acr, then maybe canon has written some excellent hardware and software based noise reduction algorithms. care to share with adobe, canon?
third, i rarely see large crops of high-iso images. i often see small crops (ie, 200px by 200px) and that’s just not a big enough crop to see banding and weird noise patterns. i also wind up seeing high iso images compressed for web use, with an accompanying statement like “omg this camera is soooooo cool.” well, great, that’s just completely unhelpful.
fourth, all this completely ignores what these files will look like spread across a 30″ album spread. that’s my next step. many 5d mark ii users are saying the noise patterns and characteristics look infinitely better in large prints compared to the 5d. so i am going to get a bunch done and report back.
fifth, and last, measuring the usefulness of this camera will also require me to assess the rumored autofocus improvements (the camera does seem more accurate, but i have done nothing to really test this) as well to determine the viability of incorporating its video capabilities into my business.
********** UPDATE *************
i am getting a far nicer noise level from the sraw files. observe the following 100 percent crops. the raw was downsized in these shots. shot at iso 6400 with an 85mm 1.2L II, at 1/80th at f/1.2.
btw, i see the infamous black dot phenomenon in the RAW file but not the sRAW file. hmmm…
Thanks for the info and for taking the time to post. Useful stuff to know. After producing the 1DS III one would think they could hopefully produce a more ‘budget’ sensor to replicate their success with the original 5D compared to the 1DS MK II.
I’ll watch your blog with interest.
Kind regards and a happy Christmas to you & your family.
Mark
Thank you very much for a solid initial review. I look forward to hearing your impressions on the printed images. I too am looking at the 30″ size. This and the hope for better high ISO performance were my primary desire to upgrade. I think it is fair to say from your assessment and others, we are looking at a 1/2 stop at best if you want to keep the resolution.
The sRAW example is interesting, and congruent with several other opinions I have read. I think this points to a separate possibility. That this is a hi-rez camera (21mp) at iso 50-1600 and a high ISO camera (sRAW) at iso 3200-12,800.
Thank you for your thoughts and review
Thanks for taking the time to make these tests and post them. As a lifelong Nikon shooter, I purchased the 5D II for landscape, commercial and architectural work. I thought I’d give Canon a try and ordered it with the 24-105(BTW I love this lens,wish Nikon had something similar). Anyway, I too did a series of tests and was also disappointed. Not only did I immediatly encounter black dots but also severe banding and sensor pattern noise. I was shooting in RAW and also using Lightroom as is my normal workflow. To be honest, I was not impressed at all with the image quality and the huge 30MB 14bit files really take up space on the computer. I also found the af focus sluggish with af points in wrong places and not enough of them. There are other minor things too like the histogram being too difficult to read outdoors in bright sun and having to press the playback button before being able to zoom in and the info menu cycling forward in one direction during playback and not enough customization with buttons. Unfortunately, I’m sending the camera back for all these reasons. I really wanted to like this camera and was planning to make a serious investment in Canon lenses, but those plans on hold for now. Not trolling, just my honest opinion.
Cheers,
Scott
I am on my way to purchasing the new 5D, so I can’t respond from personal experience, but in that first photo, in the 100% view, I can clearly make out “Honeywell” on the thermostat, and cannot do so with the 5D. That is a clear and convincing difference. Sum up that difference picture-wide, and you are talking the difference between a consumer lens and an “L” or Leica lens in resolution, ie. noticeable to the discerning eye. As for noise, generally speaking, it’s a matter of signal to noise in the sense that well exposed areas have high signal to noise ratios whereas the shadows have much lower signal to noise ratios. In other words, the bad noise is found in the shadows, not the well lit areas. Looking at the shadowy areas and black items, I notice a distinct difference between the two in favor of the 5DII. If I were a wedding photographer, I would certainly appreciate that difference, as many dramatic photos include significant areas of shadow, and a trend towards natural light is in place (in fairly poorly lit venues to boot.) While you may be satisfied with your poster sized output from the 5D, the ability to creatively crop simply cannot be underestimated. That is a huge advantage, even if you end up with the same number of pixels in the end as the 5D1.
Nonetheless, nice effort. Thanks mucho.
Richard
I’m not sure I totally agree but only time will tell. The extra .9 frames per second is a good measure for the amount of “improvements” that were introduced. I really have not noticed the extra bump in shooting speed. The video is cool but not useful to me. The self-cleaning sensor is nice but I never really had a problem with the original 5D or, 5D Classic as I prefer to say. The ISO sensitivity issue should be tested and shown in real low light situations with faces. The week I got my new 5D Mk II, I spent a few days on Capitol Hill shooting for a news outlet and was able to shoot both cameras side by side and was pretty happy with the low light capabilities of the new body. I shot up to 4000 ISO and it looked pretty good. I had to process my files with PS and ACR and have not, to this date, tried to use LR. It is also worth mentioning that I have not tried the full RAW file setting. I have only used sRAW. I am not convinced that we have been duped by Canon but am also not confident that the new camera is much more than a tease. We shall see. BTW, focusing with Live View is pretty cool if you have time and want to check critical focus at say f 1.2 or 1.4.
Thanks for the posting and have a great holiday.
I think you might be jumping the gun a bit. Software updates and firmware fixes will come along soon enough, I suspect. I see more range and detail in those Mk2 pics. I’ve seen plenty of high ISO sample pics around the web which suggests with the right post processing the images are definitely a considerable improvement. Based on your shots though, I’m not sure I’d use anything above 6400 seriously (and I don’t use 3200 on the Mk1), but they are probably usable as snaps (in very low light), and might even make good smaller prints? I’m not trolling either, I just upgraded from a 30D to a 5D Mk1 and realise what I’ve been missing, so I’ve been thinking along the lines I’d really like to upgrade again a bit sooner. I can’t afford to do that yet, but I will keep an eye on developments.
Adobe failed to get it right first time with Lightroom when the 40D was released and also the 50D (I have both cameras). It required a subsequent point release of LR, for each camera, before it could compete with DPP on IQ. in particular noise/grain was the most obvious deficiency with the early LR releases. Remember the bruhaha over the RPReview of the 50D, where they started out with a beta ACR release on then retested with the first live release. I’m afraid they should have waited for the second live release in order to be fair to the camera. I would not be surprised to find that LR support for the 5D2 improves in the next release of LR. I doubt very much then any such improvements will be documented in the public release notes.
I also find that the choice of camera profile makes a considerable difference to the results. Prior to LR 2.2, I was experimenting with the beta camera profiles and settled on “Camera Standard” as my prefered starting point. Now that LR2.2 is delivered with the production/live profiles included I am finding that “Adobe Standard” is my profile of choice and now set as my default when importing raw files.