Is canon 5D Mark II suitable for low light photography

<p>Now the 5D2</p><div></div>

<p>Very good. Now downsize the 5DII image to match the resolution of the 1DIII image and compare them again. You will now see an image that is half as noisy with the same detail as the 1DIII, in fact MORE detail than the 1DIII, because the effects of the Bayer interpolation have been taken out by down sampling.</p>

<p>There really is no argument here. The 1 series is for redundancy, better sealing, better speed, and better (and more) AF points. It's a rugged tool designed for industrial use.</p>

<p>But what you can't say is that it is the best image quality of the Canon line. At least until the Mark IVs come out.</p>

<p>Thanks David for the samples. It is strange that the noise actually looks different. Like film grain from diferent kinds of film. I thought the 5D2 was the only camera which go go to 25000 ISO, but you are correct. I have very little experience with the 1D series camera's. I think the 5D2 is as close as I need to get to a real pro camera. the 5D2 is expensive, but persnally I think with the addition of the HD video is is worth it. Espeacially if it holds its value like the 5D.</p>

<p>Its a rhetorical question, surely. And anyway, as wedding shots are so precious because you can't go back and do them again, every photographer I know tweaks their focus manually at weddings anyway, because they are shooting at very low f stops and the resulting shallow depth of field can be tricky to handle.<br>

And anyway, why would you not use flash? Its a wedding after all, not night street photography. You're not going to be creeping around getting candids...the client won't pay for those. They want a set of memorabilia to show their kids and grandkids.<br>

The latest flash units are so versatile you can just use them to brighten the shadows or use bounce to brighten the whole scene. So who needs iso3200?</p>

<p>No need to resize the images. They were both at 100% pixel level just as the sensor produces them and the same size crop was taken so that neither crop had its pixels packed tighter or looser. Its a 100% crop of exactley how much noise the sensor produces.</p>

<p>Doing it this way doesnt magnify any noise, what you see is what you get. If you view an image at 50%, the image appears to be less noisy. The fact is, the 5D doesnt produce an image of 3888x2592. So downsizing it makes no sense. You change the game then.</p>

<p>You can also review the DP Review and they state the same thing. The camera applies heavy NR. Look at there samples turned with NR off and its alot noisier.</p>

<p>Minute, it is diffeent. They pull out more Chromiance noise. Not much, but very little. It makes it more tolerable to view. The luminance noise isnt as bad as Chromiance. Thats why the D3 looks so good. Lots less Chromiance noise and the D3 tens to over expose slightly which helps subdue noise. You are right, its close enough. Its a great performer, but I couldnt stand by while a statement was made that it was leaps and bounds better in ISO performance and I know its nowhere near as good in AF as 1D3....just isnt so. Though not in the same class, understandable, but you can say its the best if it isnt.</p>

<p>The sensor produces a slightly noisier image. Thing is it has 21mp and we get these results. Sure would be nice to get a lower MP camera with a truely better noise control.</p>

<p>Stephen, I need ISO3200. I shoot weddings. I shoot candids and my clients do pay for them, and they specifically ask for the great candids moreso over the posed stuff. I also use flash almost the whole time. Sometimes, you need ISO 3200 even with flash. Bounce flash isnt used to just brighten the scene. Its mainly used to lift shadows, but done softly cause of the bounced bigger light source thus removing harsh shadows. If bounced properly and not just the tilt forward 45 degree method so many misuse.</p>

<p>I never "tweak" my focus manually. Only if I'm having trouble getting focus lock or very fine DOF and I need one very thin part in sharp focus, otherwise a wedding flow is too fast paced to manually tweak every single shot.</p>

<p>Daniel--</p>

<p>You say "(I mention the DoF because it can be razor thin with, say, an f/1.2 lens, so thin that even slight body movement forward or backward on your part could require a refocus.)"</p>

<p>Now, combining this with your 'focus & recompose' technique, don't you run the risk of the main subject falling out of focus when you recompose due to the razor thin DOF?</p>

<p>Absolutely agree with you on separating AF from shutter release. In addition to allowing one to shoot without constantly refocusing, it helps to be able to meter separately from focusing.</p>

<p>Let me know what you think about razor thin DOF + recomposing. I actually use the other AF points just b/c I'm afraid that recomposing might cause the main subject to fall out of focus when shooting at f/2.8 (certainly more of a concern for f/1.2!). But, I have to admit, I hate using any of the non-center focus points-- in the dark they continuously hunt!</p>

<p>If DOF is not a problem with recomposing, I'll certainly switch to just using the center AF point...<br>

Thanks,<br /> Rishi</p>

<p>Looking at David Amberson's 100% crops, I'm compelled to bring up an unrelated topic: look at the unnatural noise in the form of horizontal and vertical lines. That's right, not random noise as in film grain, but little vertical and horizontal 'slices' all over the place (best way I can describe it).</p>

<p>David, did you use DPP or ACR/LR to process these photos?</p>

<p>I see this type of noise in my Panasonic LX3 RAWs also when I use LR <strong>or</strong> when I use any other open-source software (RAW Photo Processor, digiKam, etc.) and select 'AHD' demosaicing. If, however, I switch to 'VNG' or 'VCDMF' demosaicing, I get much more pleasing noise rendition without these horizontal & vertical 'jaggies' or what have you. RAW Developer also uses some demosaicing algorithm that renders noise much more 'pleasing' (i.e. random).</p>

<p>Have you guys noticed this? I wonder why on earth more 'sophisticated' software such as Adobe's don't allow for more demosaicing algorithms and for us to choose which demosaicing algorithm we wish to use!</p>

<p>Thoughts?<br /> -Rishi</p>

<p>P.S. Here's a visual comparison of the noise rendition of various demosaicing algorithms:<br>

http://www3.elphel.com/importwiki?title=Demosaic_on_client_side</p>

<p>Allow me to rephrase. Compare ISO 3200 noise between the 5DII and 1DIII in an 8x12 print.</p>

<p>It is indeed changing the game when comparing apples to apples. Your pixel to pixel comparison is misleading. When comparing equal sized images, the 5DII images will always look better, both in resolution and noise.</p>

<p>And as I said, I do not use noise reduction. I personally believe that a noiser image looks better because it has more detail. Any noise reduction technique must reduce detail. I also use lightroom, so default settings are the same for all of my cameras: 0 luminance and 20 color, Which I often set to 0 and 0.</p>

<p>I also shoot weddings, and I must say that I shoot candids in low light, and I have never had to tweak the focus. The AF works fine. If it's so dark that the original 5D and a 1.2 lens can't focus, it's too dark to see. With the 5DII, same deal.</p>

<p>I'm the kind of guy that feels money is no object when it comes to photography. Forgetting money, I chose the 5D and later the 5DII with canon's L primes for weddings. I did so, never regretting it. To make up for the reliability issues, I bought three of them. I also shoot Leicas and Mamiyas. I'm very comfortable in that for my shooting style, there is no 35mm SLR kit better than than a 5DII and Canon's fast L primes. At the moment anyway. If I had to get rugged, or get two cards written at once, or shoot closeups of down-hill skiiers moving toward me at 85 MPH, then I'd compromise a touch of IQ and use the 1DsIII. Since I largely shoot studio, street, and events, the 5DII wins.</p>

<p><em>You say "(I mention the DoF because it can be razor thin with, say, an f/1.2 lens, so thin that even slight body movement forward or backward on your part could require a refocus.)"</em><br>

<em>Now, combining this with your 'focus & recompose' technique, don't you run the risk of the main subject falling out of focus when you recompose due to the razor thin DOF?</em></p>

<p>Unless you're on top of the subject or really swinging the camera on recompose, I don't think the distance from the perspective of the lens and AF module changes that much. I could be wrong for f/1.2, my fastest lens is f/1.8.</p>

<p><em>Let me know what you think about razor thin DOF + recomposing. I actually use the other AF points just b/c I'm afraid that recomposing might cause the main subject to fall out of focus when shooting at f/2.8 (certainly more of a concern for f/1.2!).</em></p>

<p>I've never seen an issue with recomposing using a 50 mm at f/1.8. However, at f/1.8 taking a slight step forward or backward, subject or photographer, will require a refocus in many situations. It all depends on subject distance and how critical a precise plane of focus is. It's surprising how easily one will shift their weight and move a bit without even thinking about it.</p>

<p>I agree that the 5DII is a great wedding camera. After all, the image's final result is a print which cant show a 100% crop. So the image must look its best viewed normal and not 100%. That said, the 5DII is amazing when viewing pictures as pictures. Canon has the right combo of NR, sharpness etc to make a final image look good.</p>

<p>The 5D2 does make very sharp, fine detailed images. What I wish my 1D3 had was a little better resolution. Its not always a big deal, but if the subject doesnt fill the frame, then the image sometimes lacks "bite" Of course, because the subject is only filling a small portion of the sensor thus only has say 1/10 of 10MP to resolve fine detail, where a higher res sensor givent the same subject placement would have a higher resolving power.</p>

<p>If I fill the frame with my subject, it can make amazing 13x19 print's even viewed at 6" away. Thats something the 5D2 doesnt have any trouble with even if the subject fills just 1/3 of the frame. The fine deatil is still largely there. It is amazing that it can be as clean as it is with that kind of resolution.</p>

<p>I hope the new specs rumored for the 1D4 are correct. Possible FF or 1.3x with 16mp and better ISO performance than the 5DII. This is what I'm waiting on to see. If Canon delivers some mediocre update to this camera, I'm jumping ship. I wanted to wait and see what Canon did with all the latest Nikon comp. before going thru the heartache of switching. I have too much money tied up in Canon gear, but at least if I see the 1D4 hasnt really changed, then I can switch to either a D3 at reduced price(maybe D4 out) or jump into a D4 if its ready by then. Either way, I win cause if the 1D4 is great, then I dont switch, if it isnt, I get a D3 which we know is great, or a D4 which I'm sure will be great. My 1D3 is good, but I want better ISO, more res. and a better LCD for god sake most of all.</p>

Hey Rishi.... I remember a while back you posted about these vertical and horizontal lines in your 5D(II?). I also chimed in by saying I get the same thing in low light with my 5D. Is the demosaicing algorithm the answer to this problem? If so, how does that fit in with all the theorizing that went on in those threads about amplifiers and where and when they were used? I might try a different algorithm on my 5D raws when I get a chance and see how it looks. Cheers.

<p><em>I hope the new specs rumored for the 1D4 are correct.</em></p>

<p>I hope the rumored specs for the 60D are correct. I know, I know, it's a lower class of camera than you guys are talking about. But if they really put in 19 point AF (lifted from the 1D module), 7.4 fps, video, and managed to improve high ISO and DR in their APS sensor, it will be one heck of a mid level camera. A true match for the D300 and D300s.</p>

<p>Since I can't really justify the budget for a 5D mkII or a 1D4, I'm hoping for a solid 60D :-)</p>

<p>For the record, while I don't have an issue with AF in low light because of technique, I would like improved area AF for BiF and airshows. The xxD AF isn't bad here, just not all it could be.</p>

<p>I got lost in here somewhere. First we seemed to be comparing the 5D II to the 1DsIII, and the next thing I know we are talking about the 1DIII, not the IDsIII. Did you mean IDsIII on those pics, David? You have them labeled as the 1DIII (1D3).</p>

<p>--Lannie</p>

<p>While slightly off the original topic I have to agree with David Amberson on the 1DIII vs the 5DII. The AF on the 1 series is significantly better than on the 5DII and the 1DIII is slightly better at high ISO as David's photos show. The real issue is what do you need to do the job. In critical situations and professional work the price difference is worth paying to give yourself the best chance of success. This does not mean that the 5DII is a poor camera - it is probably the best price / performance compromise that Canon makes (the 40D and 50D are other potentially strong competitors for a price/performance award). I own also all of Cannon's professional bodies from the F1 onwards but have not bought a digital 1 series. I do not shoot for income these days so the extra cost of the professional body is hard to justify given the shorter lifecycle of digital bodies. I have shot a number of 1D bodies at ski races and ice hockey games and the 1DIII beats the 5DII, even the 1DIIN offers pretty good performance. The point is that the 5DII is good enough for almost all shots. Set up correctly with a fast lens (300 f2.8 or 70-200 F2.8) you can get a 99.5% or better sucess rate at ski races and a 90% success rate at poorly lit indoor ice hockey games. For most of us this is more than adequate - but if you miss Herman Maier falling or Jerome Iginla scoring you may be out more than the price difference between the camera bodies. Thus for a professional sports shooter you have to have a 1 series AF system. For weddings the case is less clear and the 5DII has a better LCD (for live view)and the ability to shoot video - this may make it a better option than the 1DIII.<br>

From the few 1DsIII shots I have taken, I feel that the 5DII has better high ISO performance (e.g at 3200) but this is a non-scientific test.</p>

<p>Bernie,</p>

<p>No in those threads I was talking more of lines across the entire image... which, logically, <strong>does</strong> apply to amplifier gain variability from row to row. The noise in the AHD example from the link I pointed to above shows small horizontal and vertical lines, maybe 40 or 50 pixels in length each (wild guess)... look closely at the 200% images. I myself don't need to look very closely and can see it at 1:1.</p>

<p>This noise has nothing to do with amplifier gain variability (in my opinion)... seems to be coming from the inferior demosaicing algorithm. And Lightroom shows the same sorta noise... so I'm guess they use that algorithm or some variant of it. Don't know why, since there clearly seem to be better ones out there.</p>

<p>-Rishi</p>

<p>You need to expose right in camera to get best results with high ISO, any manipulation with software afterwards will degrade the image.</p>

<p>I would go Nikon D700 for weddings, better high iso performance than 5D11, not as many pixels but how big do you need to print for a wedding album?</p>

<p>Suddenly I understand so many people hate pixel peeping.</p>

<p>For me, pixel peeping was about per-pixel sharpness, demosaicing, and jaw-dropping detail.</p>

<p>For some, it's a great way to pretend that comparisons of image quality at 100% of cameras with vastly different resolutions is in any way valid.</p>

<p>Comparing a 3200 ISO image printed at 16x20, it is going to be painfully obvious which camera has the best noise performance. No noise reduction, no sharpening, just a plain print of the entire frame. To match print sizes, you may downsample, or print at a higher resolution. Either way, the 5DII is it. Twice the resolution means that the camera can be twice as noisy, and still equal the lower resolution camera in low light performance.</p>

<p>LOL<br>

Sure pro bodies may focus faster. But pro bodies are not required to focus in low light, or take good pics.<br>

Guess my 5D does not know it cannot focus in low light. I will not tell the 5D.<br /> It continues to focus in low light using the center focus point</p>

<p> </p>

<p>Landrum, I was speaking of the 1DIII. Sorry. The 5D2 only beats the 1Ds by only as much as the 1D3 beats the 5D2. This is because the 1Ds uses same size sensor and same pixel density, but the 5D2 utilizes and new sensor design of gapless microlenses. The 1Ds had been out a year before that tech was released. Just wait a year and see what the 1 Series does with that same tech. EWWWWWWHHHH</p>

<p>Ed, I dont know where you come up with your rationalization, but twice the resolution doesnt mean the camera can be twice as noisy and still produce better prints.</p>

<p>We've gone way off topic. My examples flat out show that sensor vs sensor without any NR and any manipulation show the 1DIII puts out a less noisier file. Not by much, but its cleaner. We can talk all day long about the resolution and yes its good, but lots of it is wiped out with the strong NR. Looking at the files just as the camera makes them, the 1D3 is cleaner...thats it. The AF is miles ahead. The 5D had better LCD and better resolution. Other than that, its not better. I dont need video. I'm a photographer, not a videographer and my brides hire proffesional video guys to do that part. If I needed the capability to print bigger than 16x20, I still wouldnt buy it cause the AF is so inferior. Having used the 1D3 for over a year and trying the 40D you can see the difference immediatley, and the 40D is better than the 5D2 in that respect. This is why if the 1D4 should be a perfect fit for me. More res.,better screen, and I'm sure they overcome any AF issues from the past. The 5D2 could quite possibly be a backup, but not main body.<br>

And BTW for those who said something about focusing in the dark, my 1D3 has focused in a gym during a pageant with almost no light. Only sillouettes were visable and I got the shots, everyone next to me using Canon XXD's, Sony's and even Nikon DXX's all gave up.</p>

<p>But not me, the very first shot in <a href="http://www.amberbrookephotography.com/2009%20Pleasant%20Grove%20Dance%20Show/">this</a> linked album was in complete darkness as you can see where the only light was my speed light, ISO 3200 and at 200mm you see how far away she was from me. Bounced flash over my shoulder into the wall behind me and ceiling above. It focused almost instantaneous, never hunted. This lady found me at a baseball game 2 months later and thanked me for that shot cause it was the only she had so far with her new grandson. It wasnt the best shot, but pretty good given the scenery and this is a 1024x768 downsized version.</p>

<p>I later earned a job from that photo and thats why the pro series 1D3 is the choice for me. I'd bet money that any other Canon(non 1 Series) would not have got that shot. Its simple, if you "need" 1 Series, get one, if you dont "need" it, the 5D2 is the next best thing and will be one hell of a camera.</p>

<p><em>Ed, I dont know where you come up with your rationalization, but twice the resolution doesnt mean the camera can be twice as noisy and still produce better prints.</em></p>

<p>Yes it does. You have to enlarge the lower resolution image more for any given print size. Which means you amplify its noise by the greater enlargement. But you cannot amplify its resolution. If in the enlargement you end up amplifying the noise to the levels of the higher resolution sensor you now have no noise advantage and a significant resolution disadvantage. This is inescapable. You cannot ignore that the final product of any camera is an image printed or viewed at a specific size, and that you must enlarge from the sensor to reach said size. The more you enlarge, the more you amplify problems.</p>

<p>Per pixel the 1D3 may be cleaner. Printed to a specific size the 5DII will prove cleaner. To do a real comparison you must upsize the 1D3 image to the 5DII dimensions, or downsize the 5DII image to the 1D3 dimensions.</p>

<p>There is a paper on the Internet which discusses this in greater depth. If I can remember or find the link I'll post it.</p>

<p>Again - I have to defend David not on resolution but on AF. Since I have the 5DII but do not own a digital 1 series the AF in low light on the 1 series is much better than the 5DII. Even my pair of old EOS1Vs can focus quicker in very low light than the 5DII. The 1DIII and 1DIIN are probably slightly better than the old 1V but it is not easy to noice and would require instrumentation. Your point Paul is that the 5DII meet your low light requirements - that does not say it is faster than the 1 series. The simplwe fact is that Canon was afraid the 5DII would take away pro body sales so they used an inferior AF system. Even the 1 series needs an AF upgrade - it is not much better than the 1V was and if you have not tried the Nikon D3 try it - the AF on this camera is notably faster than the 1 series. Don't let product ownership blind you to weaknesses otherwise manufacturers get complacent. I have NEVER owned Nikon SLRs but they currently have Canon beaten for AF. Similarly I own the 5DII but I can assure you for professional sports use the 1 series is the only choice - for amateur use the 5DII set up correctly with fast lenses is still very good. As I said in my earlier post we are talking about a few missed shots between the 5DII and 1 series. If all you take is sports shots and you must get them all buy the 1DIII. If you need a more general purpose camera the 5DII is a much better compromise (cheaper with no 1.3x crop, higher resolution, better dynamic range and a much better screen).</p>

<p>By the way Paul most professionals are required to take good pictures (and focus in low light).</p>

<p>In my experience of shooting night events (no weddings), if the subject is decently lit or if the moment can wait for you to focus and recompose, the AF of 5D/5D.2 is enough.</p>

<p>If your subject does not have enough light on it, and you don't have time to focus and recompose, you can forget about it. Of course faster lens helps, and AF assist also helps, but they can only go so far.</p>

<p> </p>

<p>So to summarize:<br>

For weddings, the 5DII beats the 1DIII in noise performance, and as long as you use the center point to focus, equals it in AF performance.</p>

<p>Low Light Winner: 5D Mark II</p>

<p>Mark Anthony Kathurima above posted a link to Jeff Ascough's comments on the subject. As he is a world class wedding photographer that specializes in low-light, no-flash candid photography, I'd say he trumps the discussion. He uses his 1 series cameras as backup since the 5D2 came out.</p>

<p>I have a similar wedding photography style, (though I use the flash crutch a bunch more than Jeff Ascough does, because let's face it, he's a master), and the 5D or 5D2 have never let me down in a low light situation. Granted, I'm using very fast primes, and I'm aware of the limitations. When working on a dark dance floor, flash or STE2 assist is a good idea, and you are always close enough for that at a reception. You can't see anything through the viewfinder anyway when it's that dark. When those situations happen, I use the red beam to aim the camera while holding it over my head. (which is easier with a lighter camera :)</p>

<p>See this series:<br>

<p>Unless you shoot unlighted midnight weddings, the 5D II will be excellent for more typical situations.</p>

<p>Dan</p>