brainiac wrote:
I am not sure it's _worse_ than the 5D in regards of shadow pushing bandwidth, but it's not enormously better. I certainly agree with you that the both 1D cameras seem to be better in this regard.
Maybe eosfun can enlighten us on how many bits the camera thinks with at each stage of image processing. I don't believe I am seeing a full 14 bit path.
I just know that "I" didn't like the detail or quality - which I know is subjective - and If I didn't like it then I won't inflict it on my clients.....
Or of course it may be my poor technique - luckily though I've sorted it and got what for me is the better tool for my work - 1 series
In my work I seldom shoot below 800, and up there the 5D2 seems to do very well, but I am now a little disconcerted about having sold my 1Ds3, for the rare occasions when I need high latitude or studio isos.
It's possible that this bandwidth bottleneck is related to the electronic constraints of handling video.
brainiac wrote:
In my work I seldom shoot below 800, and up there the 5D2 seems to do very well, but I am now a little disconcerted about having sold my 1Ds3, for the rare occasions when I need high latitude or studio isos.
It's possible that this bandwidth bottleneck is related to the electronic constraints of handling video.
I'm perhaps 50/50 as all bridals etc are lower for the best quality .... and frankly I was dismayed at the backwards step in this area... I managed well before so will do so again as I want the ability to take top quality studio shots
The low ISO performance problem isn't new. I've seen it while doing HDR shots with a 20D. The middle exposure and the over exposures lookes fine, but as soon as you went under the noise got bad real fast. I suspect (no proof, so don't flame me, just a theory) that Canon, in order to boost the high ISO performance had to sacrifice some low ISO performance. It seems that everything in photography is a trade off.
Many of you also seem to forget that not everything is shot at the "correct exposure". The correct exposure doesn't reflect the real situation in many cases. Last June I was shooting one of the chapels on the bridge in Avignon. It was very dark in the chapel with bright sunlight streaming through a small window. The "correct" exposure showed the interior as very well lit with a very bright window. The sunlight on the floor was almost lost in the "correct" ambient light.
To get the image to appear the way the chapel actually looked to the viewer's eye it was necessary to bring the exposure down. The OP was trying to convey the the great color you can get on a dreary day (just a guess Richard) and succeeded except for the noise problem.
BTW, Andi.......the histogram does not show under exposure. It showa an image where the majority of the luminance is but still within acceptable range. If it were underexposed, the graph would be going off the left edge.
I have a 5D2 and I see noise on the dark areas even at ISO 100 but show me any camera that has 20 MP, no noise on shadows and still retain the details (some other companies would prefer to do in camera NR thus damage the details and claim their fame). Of course with similar price .
I did a test with my 5Dmark2 vs 1Dmark3 both with the same exposure and same lens. ISO 100, Shot RAW and converted both with Capture One 4 with all noise reduction turned to zero I then cropped out roughly the same area of each photo then put them together in CS4, I didn't re size either image or adjust anything other then when I had them in the same file and flattened I put them back to 8bit and srgb.
Braniac, are you sure that the 1ds3 doesn't exhibit that type of shadow noise? A side by side comparison would be beneficial. One thing is for sure-you would get lower shadow noise by overexposing the shot and pulling it back one stop, or even shooting at ISO 400 @ the same exposure and pulling it back one stop. Those shadow areas are always problematic for digital cameras and that's a tricky exposure.
kazman442 wrote:
I did a test with my 5Dmark2 vs 1Dmark3 both with the same exposure and same lens. ISO 100, Shot RAW and converted both with Capture One 4 with all noise reduction turned to zero I then cropped out roughly the same area of each photo then put them together in CS4, I didn't re size either image or adjust anything other then when I had them in the same file and flattened I put them back to 8bit and srgb.
File is 1.2meg so I am linking it.
Original Size
Good comparison but the issue is whether or not the 5d2 responds differently to shadow areas than the 1ds3. Shadow noise does not behave the same way as noise in correctly exposed areas.
orangefirefish wrote:
One thing is for sure-you would get lower shadow noise by overexposing the shot and pulling it back one stop...
A ground-breaking idea, of course, but what would you suggest I do with shadows in the commoner type of scene where dynamic range far exceeds the sensor's range? Blow all my highlights I suppose. Come on fellas - you're better than this.
>Those shadow areas are always problematic for digital cameras and that's a tricky exposure.
Brainiac, I really don't get what you're saying about bandwidth, etc. Don't see how that factors in.
How did you meter the shot?
If you used Highlight tone priority, or whatever its called, it's my understanding that the camera then underexposes by a stop (or 2/3rds) then bumps up the exposure in in post processing. So if you used that, your shot *is* underexposed by one stop. It doesn't look like there is a huge dynamic range in the scene, so you probably would have been better off shooting without highlight priority.
All of this recalling from memory and speculating can easily be resolved by a side by side comparison of a 1Ds3 and 5D2. For those that own both (or at one point that owned both), I'm surprised no one has stepped up to offer a conclusive side by side set of images with the same lighting and exposure, since it would be trivial to do. The only comparison images I've seen indicate that the the low ISO deep shadow noise is about the same for both bodies (since I don't own a 1Ds3).
OK - I can see I'm not getting through here. First, HTP was switched off. This is 200 iso. Secondly, the exposure has not been adjusted in post. This is what the shadows of a 200 iso shot look like, on a good day, or a bad day, in normal circumstances. It is irrelevant that this scene lacks highlights because it was very flat. Any shadows in a 200 iso shot will look like this. It is not always possible to expose to the right, since many, or even most scenes would be clipped. The content of this shot is irrelevant. Shadows look like this at 200 iso. If you exposed to the right, you wouldn't really be shooting 200 iso, would you? If you overexpose a normal scene and get away without blowing the highlights, then you're really shooting a lower ISO. The shutter speed and aperture prove it.
Yeah Kazman...since you have both ...would you mind doing the test with some dark shadow areas ?. Even IF both camera show the similar quality I'm sure some people from the other camp (as normal) will say how crappy 1Ds Mark III is .
orangefirefish wrote:
Good comparison but the issue is whether or not the 5d2 responds differently to shadow areas than the 1ds3. Shadow noise does not behave the same way as noise in correctly exposed areas.
If you exposed to the right, you wouldn't really be shooting 200 iso, would you? If you overexpose a normal scene and get away without blowing the highlights, then you're really shooting a lower ISO. The shutter speed and aperture prove it.
Brainiac, you're worth your handle, but you missed a point. Shutter speed and aperture prove nothing about correct exposure. Meter readings don't even prove anything about correct exposure.
I know you know this, because I've read your writings. Correct exposure is giving the sensor what it needs to collect the greatest amount of dynamic range data...regardless of meter readings.
Back in the film days, a lot of us who shot Tri-X and had actually done the testing knew that it did not collect the greatest amount of dynamic range data at its rated ASA/ISO of 400. It was really 320. Many films actually needed a bit more exposure and a bit less development to capture all the range they were inherently capable of capturing.
For the digital sensor--which records light in a fundamentally different manner from film--that factor still seems to be true. It really needs a bit more exposure and a bit less "development" than plugging the ISO into a light meter might indicate. The result--if you don't--is noise in shadows where there would otherwise be detail.
Rubber Soul wrote:
So many people saying that the 1Dsmk3 image quality is preferable to the 5Dmk2's. I don't have the kind of $$$ to splurge on a 1Dsmk3, so I did the next best thing: I downloaded RAW files of those test shots at Imaging Resources. ISO 100 studio shots, I processed with DPP with no NR and no sharpening. I raised exposure by +2 EV to reveal any noise in the shadows, and saved as JPEGs. I took 100% crops off the darkest area in the entire sample, shown below.
My initial trial showed slightly more noise in the 5D2. But then I realized that DPP was activating ALO on the 5D2 (but not the 1Ds3), hence the extra noise. I had to re-do the conversion with ALO deactivated in both cases. And here are the final results:
Maybe my eyes are failing me, but I see absolutely no difference between the two. Nothing that would suggest that the 1Ds3 was better in any way. And these results seem to confirm what DXOmark had quantitatively measured: That both cameras show virtually identical performance at ISO 100. ...Show more →
Rubber Soul - maybe you just need a calibrated monitor.
Brainiac is correct on this issue. Even on these tests I see a noise difference. The 1Ds3 is better than the 5D2. Just compare the bottom third of these test shots - there are pink blobs in the 5D2 test that aren't in the 1Ds3 test. Of course the 1Ds3 should be better at more than twice the price of a 5D2.
On the bright side - if you own a 5D2, love what it does and can't see any noise problems, purple fringing, or whatever with your files, why worry about what others say? Go out and have some EOSfun (can't believe I just wrote that).
Ok, I gave it a try to see if I could do what you are saying, Exposed for the outside and included the inside that is dark, I do see noise but it's not banded that I can tell or it does not seem to have the pattern noise, See if you can see it, I am including a crop at 100%, No noise reduction was used at all, Shot RAW converted in Capture One 4, ISO 200. Ron
Sorry, not buying this post at all. (No offense to the OP) I just tested in my office and nothing. Looks the same on the 1Ds3 and 5D2. This photo looks underexposed in which case you will notice noise.
Clearly this minuscule difference in noise is going to totally ruin all attempts at successful photography with this camera. I would have a hard time finding anything any less significant to worry about in the world of photography. Sorry.
Chrono1081 wrote:
Sorry, not buying this post at all. (No offense to the OP) I just tested in my office and nothing. Looks the same on the 1Ds3 and 5D2. This photo looks underexposed in which case you will notice noise.
Cool! Someone who has both and has compared both. Can you post a side by side example please?
