Those "in-between" ISOs...

[Maleko]

...instead of jumping to 1600, I like to use 1000 for soem indoor shooting that doesnt quite need to be 1600, so i guess you coudl say I use 1/3 ISO's more so at higher ends.

I would suggest ISO 1250 instead, because it has noticably less noise than ISO 1000 for the same exposure.

<div style="CLEAR: both"]</div>

Interesting! I will remember that! cheers

[Daniel Browning]

So if I understand correctly, ISO 160 on my 5D mkII isn't necessarily a bad thing as 125 would be?

Correct.

But 200 would be better and then bump my lights down a third. Right?

I wouldn't call it "better". Think of ISO 160 as "ISO 200 with +1/3 EC". ISO 200 with +0 EC is better in that it has slightly less highlight clipping, but it's worse in that it has slightly more noise. If you use them with the same light (fixed exposure), and ignore the meter, then they both result in the same SNR and dynamic range.

[Keith B]

That settles it. ISO 100 and increase the lights 2/3.

[Chuck Lee]

They yeild the same result as shooting ISO 200 +1/3 EC. You lose 1/3 stop highlights, but gain less noise. They're really the same, not better. (In a fixed exposure, ISO 160 has lower read noise in absolute ADU, but the SNR is the same, so it's not better that way.)

If they're the same then what's the problem. I actually shoot shoot ISO 160 +.3ev on my 40D so that would be the equivalent of ISO 200 +.7ev. I find the the +1/3 tweeners( 160,320,640...) better thanshooting to the right with the normal ISOs. I started this because of the dark noise histograms provide by Peter Ruevski. I discovered this plot because of a post on dpreview where the poster was concerned with the amount of noise in a ISO 250 photo he had taken. Thats' because ISO 250 is 200 -1/3ev. It's essentially under exposed ISO 200 which increases the appereance of noise in the image. In fact, it looked worse thanISO400. Ruevski's plot confirms thisfinding.As far as "dynamic range" I can still pull down 1-2ev of blown highlights with these +1/3ev "tweeners". For me it takes the guess work out of creating raws with less noise. For Daniel, it's about milking the camera for every last drop of dynamic range available. I do not purposely over expose my raws in the hopes of increasing their dynamic range. The overhead proided by the raw format simply provides a safety net in case the photo has some over exposed areas.


I understand what you are saying Daniel, but I sometimes have to wonder whether you have real world data to back up the claims you continue to make on this subject. I have not seen personally any more overhead "dynamically" with ISO 100 images than ISO 160 images in all the time I've been post proccessing 40D RAW files. Not that what you say isn't true. I just wish my 5D had as much with the normal ISOs. I am always approaching this sublect with a Missori mentalitity. "Show me" If it works better, I'll use it.


canoli, You will find that if you do some images of the sky, say in the evening, you will see a difference in noise when shooting at ISO 100vs. ISO 160. I suggest you try it for yourself. In fact, try all the ISOs and judge for yourself whether or not it is useful. As far as combing in files, who cares? Just do a level adjustment in photoshop and watch the histogram get combed.


If you want to put the camera on ISO 200 +.3ev, do it. Otherwise, you'll find ISO 160 0ev just as good and you won't have to think about it. In fact according to Ruevski you'll get 2/3ev faster shutter and a cleaner image than shooting ISO100. According to DB you'll be giving up something in terms of dynamic range. To some that is important, to othersit's not. The art of Photography, I believe, is theoffspring ofalimited dynamic range medium.And I'd rather have cleaner images than increased DR.


Great Discussion!!


Chuck

[Daniel Browning]

I find the the +1/3 tweeners ( 160,320,640...) better than shooting to the right with the normal ISOs.

The raw files have the exact same dynamic range and SNR, so if it's really any better it can only be due to issues with the raw converter, such as the problem mentioned above with DPP. Another example would be if the raw converter didn't have a good exposure compensation tool, like Adobe.

I started this because of the dark noise histograms provide by Peter Ruevski.

His read noise plots are fine, but they only account for Noise, not Signal. (By "signal", I'm referring to light, not the average level of the read noise.) ISO 160 only looks like less noise than ISO 200 because of a digital -0.33 EC. If Signal stayed the same, then ISO 160 would truly be better. But Signal doesn't stay the same: it goes down -0.33 with the noise, so the SNR stays the same.

I discovered this plot because of a post on dpreview where the poster was concerned with the amount of noise in a ISO 250 photo he had taken. Thats' because ISO 250 is 200 -1/3ev. It's essentially under exposed ISO 200 which increases the appereance of noise in the image. In fact, it looked worse than ISO400. Ruevski's plot confirms this finding.

Agreed.

I understand what you are saying Daniel, but I sometimes have to wonder whether you have real world data to back up the claims you continue to make on this subject.

I'll post a demonstration.

I have not seen personally any more overhead "dynamically" with ISO 100 images than ISO 160 images in all the time I've been post proccessing 40D RAW files.

A 1/3 stop difference in read noise is pretty hard to detect. Most of my shots aren't even within 1/3 of my ideal exposure, there just isn't time to get it perfect.

As far as combing in files, who cares? Just do a level adjustment in photoshop and watch the histogram get combed.

The combing is fine if you have a bunch of extra (wasted) bits. It would only be harmful if Canon was doing like Sony or Nikon and building them with just the perfect amount of precision.

If you want to put the camera on ISO 200 +.3ev, do it. Otherwise, you'll find ISO 160 0ev just as good and you won't have to think about it. In fact according to Ruevski you'll get 2/3ev faster shutter and a cleaner image than shooting ISO100.

Ruevski's chart correctly shows that the read noise in ADU is lower, but what really matters is SNR, and that is not improved over ISO 100 (in fact it's 1/3 stop worse).

And I'd rather have cleaner images than increased DR.

In this case, ISO 100 provides both.

[Daniel Browning]

I'll post a demonstration.

Until I get a round tuit, here's a chart that pretty much sums up everything about tweener ISO performance. (It's for the 5D2, and other cameras are slightly different, but it gives the idea.)


http://forums.dpreview.com/forums/read.asp?forum=1032&amp;message=31378147

[clemmb]

Chuck Westfall does not answer your question but has some interesting discussion around ISO expansion on the 5D. His bottom line suggestion is to test for yourself. Its digital, test, test, test.....


http://www.digitaljournalist.org/issue0905/tech-tips.html


Mark

[canoli]

Thanks for your input folks, it is much appreciated.


Daniel, I was wondering if you could explain what "implement ISO as metadata" means. That's not your exact quote, but I think that was the gist of it.


Is it something you can explain to a more-or-less casual photographer? I love learning this stuff, but I'm not very well versed in all the terminology. If you can elaborate on that phrase a bit - without (if possible) getting too technical, I would really love to understand it. I assume not all of us know what that means, so maybe it'll be a worthwhile explanation for others too.


In any case, thanks again for your contributions.

[Daniel Browning]

Daniel, I was wondering if you could explain what "implement ISO as metadata" means.

White balance could be applied as a digital manipulation to the raw file before it's saved; instead, it is just "metadata", a recommendation for how to white balance the file in post. Currently, many ISO settings are applied as a digital manipulation to the raw file before it's saved. Metadata ISO would change that to be just a recommendation for how to push the file in post. It's very similar to how HTP works now.

Many camera manufacturers implement some ISO settings by digitally manipulating the raw file in the camera. This results in highlights that are blown needlessly (two full stops in the case of ISO 6400), larger file sizes (45 MB instead of 20 MB for the 5D2), and increases the precision needed to prevent quantization error.

There are no benefits to applying gain digitally in the camera, so those ISO settings should instead be implemented as metadata. Like white balance, the metadata ISO would only recommend a certain action during raw conversion.


Hope that helps.

[canoli]

Definitely helps Daniel, thanks very much.


I always thought of the ISO setting as a physical thing, sending more (or less) electrical current to the sensor, changing its sensitivity, its gain. That's what you mean when you say -

Currently, many ISO settings are applied as a digital manipulation to the raw file before it's saved.

So the capture is defined (partly) by the ISO setting - it is"hard-wired" into the result so to speak, right?


But you're saying manufacturers could forgo that practice completely and instead write the ISO setting into the metadata. Interesting.


I can't help thinking there must be some reason they went the way they did. It has to be more complicated and more expensive to build an auxilliary amp or dedicated electrical pathways to change ISO. Isn't there some benefit to having it "hard-wired" into the capture?