extended ISO, "H", 3200?

RyanLilly · May 21, 2008

Can anyone explain what exactly canons extended ISO is, or how it is accomplished. My assumption would be that it is probably an increase in sensor sensitivity coupled with some sort of processing, witch is why is is designated as "H" rather than a true 3200(although exif data reads 3200).
Basically, If I need to use "H", is there actually an increase in sensetivity, or is it more or less just software, and if so should I just shoot at 1600, and pull it up a stop in post?

Thanks,
Ryan

Alfred D. · May 22, 2008

I would test drive both M.O.s, and put the results up side-by-side on the screen or even print them to see if there's a difference, and if so, what! In my real life situation, with my gear.
Then I would know.
You should try it.

Rhubarb · May 22, 2008

This is how it is for Nikon's and I am going to guess that Canon is the same/similar.

The ISO setting on your camera represents the International Standards Organisation's' (ISO) standard for light sensitivity.

So for example ISO 800 should have the same sensitivity to light regardless of the type of digital sensor or type of film. ISO 800 is ISO 800, so to speak.

The reason Nikon (and I assume) Canon use a 'H' or boost setting for very high ISO's is that when used at this setting the light sensitivity isn't of true ISO specification. In other words think of ISO 3200 on your camera as about ISO 3200, not precisely ISO 3200.

This can be an important consideration when trying to calculate a correct exposure, and for example you need 1 more stop of light, going from ISO 1600 to H (about ISO 3200) isn't going to equate to exactly one stop; as going from ISO 800 to 1600 would be.

So putting your camera to the ISO 'H' setting does indeed increase the sensitivity of your sensor, however it isn't exactly ISO 3200.

Note: Of course some cameras, such as the Nikon D3, do have a true ISO 3200 setting, I just used these numbers as they apply directly to your situation and camera.

Seefutlung · May 22, 2008

I shoot a lot of H/S sports at night. I have used Canon's ISO 3200/H with both a 20D and a 5D.

Increasing a stop via post processing is the absolutely worst thing you can do IQ wise. I have found that improper exposure will generate more noise than increasing the ISO.

An underexposed ISO 1600 is much more noisey than a properly exposed ISO 3200.

If you use 3200 ... make sure you has a noise reduction program like Noise Ninja.

Gary

goodoneian · May 22, 2008

the "H" stands for high 1, because 3200 is 1 ev above iso 1600. that's my understanding of H's meaning at least

RyanLilly · May 22, 2008

OK, cool. I was just worried that it was just a software thing rather than the sensor sensitivity.

Garbz · May 22, 2008

As far as I understand it it is. You can only push sensor voltages or ADC sensitivities so far.

I always believed that H1.0 (ISO3200) is just ISO1600 pushed a stop in software. I can guarantee that it has nothing to do with approximation and from brightness tests it is in fact 1 EV higher than ISO1600, and not some approximation.

To backup my theory, noise increases in a non-linear fashion when going from ISO100 to ISO1600 as is expected with a system where the physical operating conditions of the sensor is being changed. However there is a MASSIVE jump in noise between ISO1600 and H1.0 which looks equivalent to simply pushing the exposure by 1 stop in software.

My guess is if it were true ISO3200 they'd simply call it that.

Alfred D. · May 22, 2008

Garbz said:
My guess is if it were true ISO3200 they'd simply call it that.

Agreed.
So "H" isn't true ISO 3200.

Socrates · May 22, 2008

Garbz said:
As far as I understand it it is. You can only push sensor voltages or ADC sensitivities so far.

I always believed that H1.0 (ISO3200) is just ISO1600 pushed a stop in software. I can guarantee that it has nothing to do with approximation and from brightness tests it is in fact 1 EV higher than ISO1600, and not some approximation.

To backup my theory, noise increases in a non-linear fashion when going from ISO100 to ISO1600 as is expected with a system where the physical operating conditions of the sensor is being changed. However there is a MASSIVE jump in noise between ISO1600 and H1.0 which looks equivalent to simply pushing the exposure by 1 stop in software.

My guess is if it were true ISO3200 they'd simply call it that.

The "H" stands for High Noise?

Village Idiot · May 22, 2008

Garbz said:
As far as I understand it it is. You can only push sensor voltages or ADC sensitivities so far.

I always believed that H1.0 (ISO3200) is just ISO1600 pushed a stop in software. I can guarantee that it has nothing to do with approximation and from brightness tests it is in fact 1 EV higher than ISO1600, and not some approximation.

To backup my theory, noise increases in a non-linear fashion when going from ISO100 to ISO1600 as is expected with a system where the physical operating conditions of the sensor is being changed. However there is a MASSIVE jump in noise between ISO1600 and H1.0 which looks equivalent to simply pushing the exposure by 1 stop in software.

My guess is if it were true ISO3200 they'd simply call it that.

Correct.

That's funny too. There are so many wrong answers before this post.

Even ISO 50 on some cameras is a function of software manipulation and not a true ISO 50.

Garbz · May 22, 2008

Socrates said:
The "H" stands for High Noise?

Most definitely! :lol:

goodoneian · May 22, 2008

so on a d300, iso LOW 1 is not iso 100, but more of a software manipulation?

Garbz · May 23, 2008

Yes the D300 has a native low ISO resolution of 200. From what I have read ISO100 is achieved not by changing the sensor but by changing the ADC's input sensitivity. The sensor captures the same amount of light in ISO200 as ISO100, and you may find image quality is actually higher in ISO200 with a ND1 filter, than at ISO100.

Village Idiot · May 23, 2008

Garbz said:
Yes the D300 has a native low ISO resolution of 200. From what I have read ISO100 is achieved not by changing the sensor but by changing the ADC's input sensitivity. The sensor captures the same amount of light in ISO200 as ISO100, and you may find image quality is actually higher in ISO200 with a ND1 filter, than at ISO100.

In fact, there was an article I read, with a provided chart, that said higher native ISO's could provide better IQ than lower ones.

This is of course on a camera that can do ISO in 1/3 stops and it's the next highest ISO, not 640 vs. 100. More like 140 (I think this is the next 1/3 of a stop...could be wrong) was better than 100, 260, 200 etc...

RyanLilly · May 23, 2008

Garbz said:
As far as I understand it it is. You can only push sensor voltages or ADC sensitivities so far.

I always believed that H1.0 (ISO3200) is just ISO1600 pushed a stop in software. I can guarantee that it has nothing to do with approximation and from brightness tests it is in fact 1 EV higher than ISO1600, and not some approximation.

To backup my theory, noise increases in a non-linear fashion when going from ISO100 to ISO1600 as is expected with a system where the physical operating conditions of the sensor is being changed. However there is a MASSIVE jump in noise between ISO1600 and H1.0 which looks equivalent to simply pushing the exposure by 1 stop in software.

My guess is if it were true ISO3200 they'd simply call it that.

Thanks for clearing that up. I was thinking that it was probably just processed, because the color noise, and brightness/contrast just don't look normal, I took a few shots with it just to see, and I thought "they should call this setting 'camera phone.'
Now I guess I'll have to figure out if the camera does it better or if I can do it better in photoshop(although I probably wouldn't really want to use either anyway) I'm betting on some photoshop and noise ninja.