|
|
Forums:
Climbing Disciplines:
Climbing Photography:
Re: [Rudmin] Is high ISO the same as under exposure?:
Edit Log
|
|
pico23
Apr 10, 2010, 6:52 AM
Views: 12166
Registered: Mar 14, 2003
Posts: 2378
|
Rudmin wrote: I find that when shooting in low light, I am always making the trade off between blurry photos or noisy photos. One thing I have realized, is that if I use a low ISO and underexpose a photo (fast shutter speed) and then ramp up the exposure of the RAW file in post production, it looks roughly the same (in terms of noisiness) as if I take the same photo with the same fast shutter speed and a high ISO. Is there actually any difference? So is changing the ISO before you take a photo effectively the same as tweaking the exposure after you take a photo? I ask because I have to go through menus to get to the ISO setting and I frequently forget to change it to what I need. It would be easier if I could just leave it on 100 and then just underexpose shots rather than change the ISO setting
First off, NO. WHile raising the ISO is merely increasing power to the sensor to gain sensitivity, there is a lot going on in the imaging pipeline that isn't going on in the RAW converter.
For instance, how can say the D200 and D40X have vastly different image quality EVEN IN RAW if both use the same sensor if a lot of processing isn't happening to RAW images. It can't. It's why Pentax, Nikon and Sony all used Sony's 10MP sensor but none had the same RAW IQ.
The D200/K10D/D40X were all CCDs but CMOS is the new fangled wonder sensor because it has some advantages, including on the sensor noise reduction. So the CMOS is doing reduction before it even enters the imaging processor which then puts your camera makers stamp on it.
After extensive testing with both under exposing and pushing the photo in RAW and exposing at a proper ISO all the way right (to the point my highlights DON'T CLIP but almost clip), I've found exposing to the right is ideal. What you describe as your process would be exposing to the left.
I was shooting hockey a few years ago for an AHL team. Canon shooters would always contact me and ask, "how are your shots so clean, everything I heard about Pentax says I have like a 5 stop advantage" (ok I'm joking on the 5 stops, but they would say 2 stops)
Actually it was about 1/3 stop (but even that was offset by the fact that Canon used more NR even on RAW images, leaving less detail for me to work with), viral marketing is awesome but rarely true, and I made up for it by exposing to the right, rather than crossing my fingers and doing the under exposure dance. Because I exposed properly, I actually reversed their 1/3 stop advantage and probably gave myself 1/2 stop and had significantly more detail to work with as well.
Finally, the higher the ISO is, the lower the exposure latitude, so basically at 100 you might be able to play around 3/4 of a stop in post processing, but at 1600 I found that I had no more than 1/3 stop (really 1/4) to get usuable/printable shots. Fortunately my stuff was only getting used for cards, web, and small low color photos.
Here is why...
Noise in most images at high (or low) iso is in the shadows. Therefore, high key scenes like say hockey will not have noise unless you f' up the exposure. So I started bumping my ISO to from 1000 to 1600 and noticed noise was very similar, only because I was no longer pushing the images in Light Room, I was now getting cleaner shadows. So my images actually looked better at 1600 than 1000 or 1250.
Then I noticed it I opened up a bit more or dropped the shutter to the point that the brightest highlights clipped just barely, the rest of the highlights were pushed right (histogram) I was seeing less and less dark mid tone and shadow noise.
So essentially (barely) overexposing at high ISO is really the key to reducing shadow noise, which is generally the only noise you see.
After that I run a selective noise reduction using Nik Dfine, which I target ONLY the shadow colors. Reds, blacks, dark purples. The ice, mid tone skin, white helmets, etc don't get any NR applied.
Finally, not all sensors have linear IQ dropoffs with ISO. An example that comes to mind K10D at ISO 400 takes a big hit vs it's competion, yet at 640 and 800 it actually improves (has better IQ) than the same competition. So don't be so sure that shooting at 1250 will yield better results than shooting at 1600 or 2500.
(This post was edited by pico23 on Apr 10, 2010, 6:53 AM)
|
|
|
Edit Log:
|
|
Post edited by pico23
() on Apr 10, 2010, 6:53 AM
|
|
|
|
|
|
