12800 ISO is NOT the same as 3200 ISO with +2 stop push PP

I set out to say it wasn't the same but after all variables were taken into account, it appears I was wrong... Originally I'd written this but in the first case it just appears to be a lighting difference, on the next page just white balance. So nothing to see here now but if you want to see how I ended up confirming what I was denying, read on:

People keep writing that you don't need 3200 ISO on a 1600 ISO camera as you can just boost it 1 stop in post processing, same for 6400 and 12800 on the 500D/50D they say don't use them, just shoot 3200 and push 1 or 2 stops. I call BS on this and did a little test earlier this month to show it. Not scientific but enough to show me that they don't just put the extended ISO ratings on the cameras for the marketing department.

3200 +2 stops pushed post processing / 12800 +0 both 135mm f/2L hand held:
http://www.flickr.com/​photos/hpulley/4194038​770/
http://www.flickr.com/​photos/hpulley/4193281​681/

The 2 pictures look like 1 long picture, can you put a space between them?

There... better?

I don't think anyone said it was as good as native ISO, but when that is all you got, that is all you got. My camera caps at 1600. I'd love 6400, but I don't have it as an option. So sometimes I push.

Bsically, Since you underexposed 2 stops on the 3200 shot, you clipped the shadows. There is no detail to bring back except black.

Yep, lol. Thanks... that's quite interesting... I may have to play with this at home.

Thinking out loud -- one reason why the 2 pictures don't "match", if you clipped the shadows in the 3200 shot, pushing it wont bring back detail.


And I assume this was shot in RAW and then pushed?

I did a similar experiment with my old 1D by pushing a 1600 shot one stop in DPP vs shooting 3200 (expanded H in camera).

The results were 100% identical.

I deleted the photos so you'll have to take my word for it.

tkbslc wrote in post #9282729
Bsically, Since you underexposed 2 stops on the 3200 shot, you clipped the shadows. There is no detail to bring back except black.

I agree with this guy.

hpulley wrote in post #9282683
People keep writing that you don't need 3200 ISO on a 1600 ISO camera as you can just boost it 1 stop in post processing, ....

Maybe I'm misunderstanding what your doing, but I believe that what is being said is that two cameras both with native ISO1600 and one camera having expanded ISO3200 (not native) That when you push/pull the ISO 1600 camera to ISO3200 you'll get the same result as a camera with expanded ISO3200. Hardware amplification will beat software amplification because you are not loosing the DR as you do pushing/pulling.

1. The illumination was obviously changed between the two shots, thus the comparison is worthless.

2. Which camera is this? 7D?

3. Was it raw or in-camera JPEG?

I wonder if the light has anything to do with what you can see - THe light from what looks like a TV seems much brighter - In order to test this you need to be in a very controlled environment.

T1i, shot in Raw, pushed post, not scientific so I may have blown this one. I will do a controlled version. Just read it in a "XSi vs. T1i" thread and in my experience it is not quite the same to do a post process push.

Regardless of the iso rating, a correctly exposed image will always have less noise than an underexposed image which has been pushed in software. At higher iso ratings the comparison is more noticable, but getting it right in the camera will always be better.

PhotosbyKev wrote in post #9282952
Regardless of the iso rating, a correctly exposed image will always have less noise than an underexposed image which has been pushed in software. At higher iso ratings the comparison is more noticable, but getting it right in the camera will always be better.

Sorry, this is not true, if the "correct exposure" is achieved by fake ISOs. The noise will be identical with the fake ISO, but the highlights get clipped.

hpulley wrote in post #9282933
T1i, shot in Raw, pushed post, not scientific so I may have blown this one

If you repeat the experiment in well-controlled setting (note: watch out even for you body posture behind the tripod, I am serious), then pls do that with ISO 1600 and anything higher. Already 3200 of the T1i is fake. Using 12800 reduces the dynamic range by THREE stops compared to 1600; for example in the scenery you posted the lights on the tree and their "radiations" get clipped.

I still don't see what this is all about, because you aren't going to be pushing 2-3 stops if you actually have ISO 12800 or 6400 as an option. This is a bailout if you need higher ISO than your camera can do.

tkbslc wrote in post #9283080
I still don't see what this is all about, because you aren't going to be pushing 2-3 stops if you actually have ISO 12800 or 6400 as an option. This is a bailout if you need higher ISO than your camera can do.

The point is, that your camera, i.e. the hardware does not have those high ISOs. For example the T1i goes only up to 1600; everything higher is pushing. If you are working with raw files, you can do that pushing later, under more control, for example without clipping away the highlights.