[UnknownVT]

When it's a single block of colour like that, it might be useful to express the difference mathematically using something like standard deviation on the different colour channels.

I think the point is whether one can actually see the difference - that is visually -
I have already said that there may well be measurable differences but visually I could not see much.

Well, if you pull any of the shadows up in the RAW conversion process, for example, you most certainly will see the difference in the resulting jpegs, as well as prints if you're printing large. The noise doesn't disappear when converting to jpeg.

With jpegs straight from the camera (from any ISO), you don't have all that much available shadow detail anyways, so you're not really going to see the difference there.

Thank you again for the explanation - so using ISO100 RAW one has more dynamic range - especially when bringing out shadow details (like push processing?) - then the difference may be seen -
so it is in RAW and when pushing the shadows up -
a normal exposure and conversion is unlikely to show mush difference -
and a straight out of the camera JPG is not going to show any visible difference either
(as has been shown in my experience with test samples from IR and DPR).

However this kind of begs the question whether ISO100 pushed to ISO200
would actually show less noise visibly than native ISO200?

---

Last edited by UnknownVT; 02-08-2011 at 11:23 AM.

[KxBlaze]

OK now I'm confused -
unfortunately I could not find that thread using various search term variations.

I could understand er1kksen saying one may see this in post processing (RAW?)
but will not see the difference when converted to JPGs - and since we all look at JPGs -
that's the reason I could not see much difference from all those 100% crops of test images from dpReview and Imaging-Resource.

However what you are saying is that someone demonstrated this with what had to be JPGs with a plain blue sky.

Now there are plenty of examples of plain blue areas in the DPR and IR test images, and I could not see much difference in the noise - again I am quite willing to say I may not be looking at the right places - but I also have been studying K-x images for quite a while, and so far none of the 100% crop or full sized images show much difference to me.

Here are 100% crops of all the Blue Patches from the Imaging-Resource.com full sized Multi Target

There probably are measurable differences - but I am not seeing that much difference visually even with these 100% crops from the full-sized test images.

Found it. It's the post by Kiwi Jono. It will not blow you away but I could tell the difference and again it is at 100% crop (which never use).

https://www.pentaxforums.com/forums/pentax-dslr-discussion/127350-iso-100-k-x-k-r.html

[UnknownVT]

Found it. It's the post by Kiwi Jono. It will not blow you away but I could tell the difference and again it is at 100% crop (which never use).

https://www.pentaxforums.com/forums/pentax-dslr-discussion/127350-iso-100-k-x-k-r.html

Thank you so much for tracking this down.

Yes, you are right I can see the difference at 100% crop -
and you are also right I was not blown away -

However just for fairness I took those crops and put them on a single image so one can zoom/enlarge them at the same rate

- at about 130% it becomes more obvious that the ISO200 has more noise.

I did the same thing with the blue sky test images from Imaging-Resource.com
Far-Field Test

this one takes closer to zooming at about 140% to be come more obvious - but I can just see it at 100%.

To me for these samples there doesn't seem to be any real practical difference for any end result - especially since we are pixel peeping at 100%
so to me the difference isn't really big, and kind of moot for any real end results.

However what er1kksen is saying that there is a lower noise floor especially in the shadows when post processing in RAW -
seems like a worthwhile gain - if one is doing push processing -
but then perhaps the scene could have been shot at ISO200
and not need any pushing?

[Tranzistors]

If you are making HDR image from single raw file, pushing comes quite handy.
Also, with ISO 100 you get one stop ND filter for those flowing water pictures.

[UnknownVT]

If you are making HDR image from single raw file, pushing comes quite handy.
Also, with ISO 100 you get one stop ND filter for those flowing water pictures.

Yes, I can see push processing being benefited and have said so.

Not by any stretch of the imagination arguing against ISO100 -
it's the big difference in noise between ISO100 and ISO200 in the 100% and full sized test images that I could not see.

[Tranzistors]

A while ago someone (Falk Lumo?) posted a photo of ISO 5 (or something like that). It was made using iso stitching. The same picture was taken like 20 times and then stitched, making its' ISO equivalent of 5. The difference was just wow. The problem is, I can't find the picture nor any pointers of how it done exactly. I begin to think I dreamt that up.
Anyway, the point is, maybe the difference between each step in sensitivity is small, but they still count when put together.

[er1kksen]

However what er1kksen is saying that there is a lower noise floor especially in the shadows when post processing in RAW -
seems like a worthwhile gain - if one is doing push processing -
but then perhaps the scene could have been shot at ISO200
and not need any pushing?

That is essentially correct, but what most people do when they bring up shadow information is not "push processing," where the whole image is brightened; rather it is where you take an image that has been correctly exposed for the midtones and highlights and just bring up the shadows, even only selected shadow areas, in order to bring the visible brightness range more into line with what an observer with organic eyes (rather than a silicon sensor) might see.

An extreme example would be shooting into the sun (which I happen to do often). There are some very bright, high-intensity highlights, naturally, and when one exposes to preserve those highlights, the rest of the scene, even the midtones, is often underexposed as a result. In post one would then bring up the tones of the rest of the image (which were underexposed in the original exposure) while retaining the information in the highlights. Having cleaner information in those underexposed midtones and shadows results in a cleaner image overall.

Notably, ISO 200 has marginally greater range in the highlights than ISO 100, but the extra shadow range in the ISO 100 files is sufficient to offset that difference (and then some).

This extra shadow range comes in handy in all sorts of circumstances. In some cases where traditionally I would have wanted to use fill-flash to lighten up a shadow area while shooting, it's now possible to forego the fill-flash and simply brighten the areas in post with no loss in image quality.

[UnknownVT]

That is essentially correct, but what most people do when they bring up shadow information is not "push processing," where the whole image is brightened; rather it is where you take an image that has been correctly exposed for the midtones and highlights and just bring up the shadows, even only selected shadow areas, in order to bring the visible brightness range more into line with what an observer with organic eyes (rather than a silicon sensor) might see.
Notably, ISO 200 has marginally greater range in the highlights than ISO 100, but the extra shadow range in the ISO 100 files is sufficient to offset that difference (and then some).
This extra shadow range comes in handy in all sorts of circumstances. In some cases where traditionally I would have wanted to use fill-flash to lighten up a shadow area while shooting, it's now possible to forego the fill-flash and simply brighten the areas in post with no loss in image quality.

Thank you again -
this is what I actually have to do
- a lot -
but unfortunately it's at the other end of the sensitivity range -
more like ISO5000 -
and when I do bring up the shadows it can look pretty cruddy.

Haven't tried it -
but how well does the K-x shadow correction work?

[er1kksen]

Yeah, up that high on the scale there aren't any cameras I'm aware of that can make it look good.

[UnknownVT]

Yeah, up that high on the scale there aren't any cameras I'm aware of that can make it look good.

The K-x may be an exception -

The hat peak puts the face in the shade - post process to bring up the facial details.
Highlight details like the hands were sacrificed because of the overall +2/3 stop compensation at shooting.

ISO5000, f/4.5, 1/15, 43mm.

[er1kksen]

Well, I suppose it depends on what you mean by "good." Not that that's a bad photo, but it is pretty noisy, seeing as that is the topic of discussion. A D3s wouldn't have given you a clean file there either, however, so that's not really such a mark against the beloved K-x.

[UnknownVT]

Well, I suppose it depends on what you mean by "good." Not that that's a bad photo, but it is pretty noisy, seeing as that is the topic of discussion. A D3s wouldn't have given you a clean file there either,

Good grief!

Of course it'd be noisy compared to ISO100 or ISO200 -
after all this was ISO5000 -
some ~ 5 2/3 stops increase in sensitivity!

@ f/4.5, 1/15
is LV (light Value) = ~2.3
that's about 2/3 foot-candles or 7 lux
this is about 2/3 level of typical city night street lighting.

So at ISO100 this would have been f/4.5 (same aperture) @ 3 seconds,
which is beyond my capabilities for any hand-held shot -
not to mention subject movement.

The photo was also a pretty highly compressed JPG -
to keep bandwidth low/loading time short.

However - how's this version?

[noaslplease]

I'm not sure if ISO 100 in derived from ISO 200, or whatnot, but you guys need to remember that the lower the ISO, the longer shutter you need, which is good for longer exposures in sunlight if you don't have an ND filter, since ISO 200 attracts twice the light as ISO 100.

It's not all about the noise :P

[er1kksen]

Good grief!

Of course it'd be noisy compared to ISO100 or ISO200 -
after all this was ISO5000 -
some ~ 5 2/3 stops increase in sensitivity!

The photo was also a pretty highly compressed JPG -
to keep bandwidth low/loading time short.

However - how's this version?

Hey, don't get the wrong idea, I'm not implying that there's any real way you could have gotten a "clean" shot. Like I said, a D3s wouldn't give you a clean file at ISO 5000. A K-x might be "cleaner than a lot of other cameras" at ISO 5000, but NO camera is going to give you a noiseless file at that sensitivity.

Like I said, not at all a mark against the K-x.

That's a pretty good NR job, I've gotta say. Is it from a single program or did you go in with layers to selectively smooth different areas?

[jlaubza]

I can actually see a small bit noise in the iso 200 images of the blue test patches above. Does it matter? To me, no but to those infected by the obsessional virus, it may be bothersome. If I need to, a small bit of noise reduction will clean up the minor graininess.

A lot of my shots are in any case not shot at 100 or 200 iso conditions - I use my K-X's high iso capabilities a lot.

The other question - does highlight correction make a difference? For me, yes it did, negatively. Highlighted areas started to look grey so I have stopped using it. The Shadow correction function though makes a small but sometimes useful difference by lightening up some shadow areas.

Regarding highlights, if in doubt, use raw and underexpose - you protect the highlights and can lighten shadows in PP work.