extended ISO, "H", 3200?

Seefutlung · May 23, 2008

ISO w/ No Noise Reduction

RyanLilly · May 23, 2008

thats clean for 3200,hell, thats just clean in general, I'll do some more looking in to this and testing, thanks for the feedback.
Faster shutter speeds do help a lot with the noise, long shutter speeds that I've been taking recently are not so great with noise in dark areas.

Garbz · May 23, 2008

No it's not. It just looks clean because you're not counting pixels. That looks like a standard print size to me, and what a lot of people don't realise is that while this picture probably looks like a dogs breakfast when zoomed to 100% if you look at it on the whole (as you should) noise becomes much less of an issue.

D200 a camera known for it's horrible noise: ISO3200, Noise Ninja with processing set to -5 so not very strong NR at all lit by a single fluro in the room next door:

RyanLilly · May 23, 2008

Garbz said:
No it's not. It just looks clean because you're not counting pixels. That looks like a standard print size to me, and what a lot of people don't realise is that while this picture probably looks like a dogs breakfast when zoomed to 100% if you look at it on the whole (as you should) noise becomes much less of an issue.

Right, Final product is all that should be considered. I have no problem shooting high ISO when the final products are small prints, or web use, because even though noise may be quite high compared to ISO 100, its a moot point, good enough, is good enough, and lower noise wont really change much

keith204 · May 23, 2008

Bottom line. Do what you need to do.

I shoot 600-800 pictures per race. 3 races per weekend. That's a lot of post processing time. Customers rarely order anything larger than a 8x10. So, at night I don't hesitate one bit to shoot at 3200. The 8x10 results of a properly exposed 3200 are great for racing photography. (40D here)

I shoot at 3200, and don't PP any of my racing photos. If I like them, I keep them, if not, I delete. Just how it's gotta be when taking so many pictures...gotta get it right before it hits the CF card.

RyanLilly · May 23, 2008

Hey Keith, If you get a chance, could you post a comparison of a shot from your 40D @ 3200, and one from your 20D @ "H" Just curiosity really.

The 40D has 6400 as "H" right? I wonder how noise compares between the two "H" settings.

Seefutlung · May 23, 2008

Garbz said:
No it's not. It just looks clean because you're not counting pixels. That looks like a standard print size to me, and what a lot of people don't realise is that while this picture probably looks like a dogs breakfast when zoomed to 100% if you look at it on the whole (as you should) noise becomes much less of an issue.

D200 a camera known for it's horrible noise: ISO3200, Noise Ninja with processing set to -5 so not very strong NR at all lit by a single fluro in the room next door:

Another ISO 3200 with No Noise Reduction Applied

Dude, not to be argumentative, but I don't understand where you are coming from. My 3200 w/out NR looks better than your 3200 w/NR (period). While yes it is very difficult to make accurate comparisons of images on the internet due to different monitors and compressions ... but some basic comparisons are valid ... in this case, my non-NR image taken with a 20D (old technology) looks superior (all things considered) to the D200 image (top) with NR applied.

My image was in direct response to the OP's question of underexposing and pulling the image up a stop in post. My initial reply was that a properly exposed 3200 is better than an under exposed 1600 ... I supplied images to support my statements.

I agree the image looks like hell at 100% crop ... and at a 200% crop it even looks worse ... so what's your point? That this can't be enlarged to a poster sized print or even an 11x14 without losing IQ??? Well duh, of course, this is after all ISO 3200.

My point, as stated previously, was that proper exposure goes a long way to minimize high ISO noise. So, when I go to print ... man, what a nice starting point in which to apply NR ... all due to proper exposure and a CMOS sensor.

Gary

PS- There are many opinions bantered about on the internet, I recommend that you look at the photosites of those offering their opinions as a qualifier for said opinion.
G

RyanLilly · May 23, 2008

Seefutlung said:
...My image was in direct response to the OP's question of underexposing and pulling the image up a stop in post. My initial reply was that a properly exposed 3200 is better than an under exposed 1600...

Just to clarify, the original question was, Is "H" actually in increase in sensor sensitivity, or is it just 1600 with in camera processing applied, or perhaps a combination of the two?

I'm sure that there is some processing applied otherwise they would just call it 3200, and if there is an increase in sensitivity, how much? How much of "H" is created with hardware and how much is software? The next question is, If it is just software, Does the camera do the best job reasonably possible, or can better results be obtained through post processing?

Some of these things have been answered, but I think that some testing will probably be required to know for sure.

Thanks.

manaheim · May 23, 2008

This is awesome guys... I've been playing with this a lot recently since the D300 handles higher ISO way better than my D100. (D100 was bad bad bad)

More and more I think that my problems have been how I've exposed, however.

What I've learned a bit is to overexpose the higher ISO slightly, and that has definitely had a huge impact... but I wonder if there is more?

Any specifics on your process would be really appreciated.

Thanks, guys. Awesome discussion.

Garbz · May 23, 2008

Seefutlung I'm not arguing anything against your point. Your 20D photo looks fantastic. It is widely known that at that time (2 years ago) Canon produced better noise results than Nikon. I was just pointing out the fact that the D70 and D200, some of the cameras considered to have amongst the worst high ISO noise performance of any prosumer DSLR, can still produce 100% usable images.

Your photos look much cleaner than mine. Mine actively required a tiny tiny bit of NR to clean it up, and looks quite a bit worse, but then I still prefer to have the image with noise, than to simply not have it. And that was my point, really just re-enforcing yours.

Plus given the crap lighting quality even with a 5D or D300 that picture I posted would look average :lol:

Manaheim. Why over exposure the High ISO when you can just drop the ISO by 1/3rd or 2/3rds of a stop and probably get the same noise result but with less work after?

Seefutlung · May 23, 2008

Okay ... sometimes the internet can be a tough place to discuss fine points.

goodoneian · May 23, 2008

RyanLilly said:
Hey Keith, If you get a chance, could you post a comparison of a shot from your 40D @ 3200, and one from your 20D @ "H" Just curiosity really.

The 40D has 6400 as "H" right? I wonder how noise compares between the two "H" settings.

i don't think the 40d has an iso 6400 equivalent, and i think 3200 is the "H" on it

manaheim · May 25, 2008

Garbz said:
Manaheim. Why over exposure the High ISO when you can just drop the ISO by 1/3rd or 2/3rds of a stop and probably get the same noise result but with less work after?

Somewhere on this board there was a lengthy thread about it. It suggested that you are better off slightly over-exposing the subject a bit in a higher ISO... shooting in RAW, of course. It said that properly exposing the subject would result in noise... overexposing it would still have noise in the shot, but not on the subject.

I tried this at a circus (which is hell to shoot at, so I didn't get it quite perfect), but this one particular shot seemed to be a great example of this actually working...