[Big G]

Hi John,

Thanks for you continued excellent contributions, appreciate getting information nice and early. I must ask the same question regarding the exposure, I've downloaded the RAW file of this portrait and it does indeed look over exposed to me too (not basing this on your screenshot above). Some of the highlights are showing as blown on the face, but Recovery is pulling those back as expected.

The only way I can replicate results very similar to yours is to pull down by 1 EV and run Denoise on that TIFF file. I would be interested to hear your comments? I'm not taking away from your samples, the high ISO is superb on this camera, but it's interesting that two of us think the same thing about the original RAW file.

EDIT:

To produce more or less the same visual impression in Lightroom 3 I have to reduce the exposure of the RAW file by -1. This should kill some of the noise. But anyway, still impressive.

Hah, that's exactly what I've found .

[JohnBee]

To produce more or less the same visual impression in Lightroom 3 I have to reduce the exposure of the RAW file by -1. This should kill some of the noise. But anyway, still impressive.

RPP display's images exactly as they are captured. Which is substantially different from other RAW developers. Is also means that the RGB histograms in RPP and the camera are identical over the entire range.

[Recercare]

It is very impressive, but I notice that we're always looking at close-ups. In other words, pics with large uniform areas (the skin). I would like to see a high iso pic with many small edges and details.......for example, a crowd of people.

Anyway, have you made up your mind about K-r vs K-5 when it comes to noise ?

Kind regards
.lars

[Adam]

Yup that's some ownage right there.

Adam
PentaxForums.com Webmaster (Site Usage Guide | Site Help | My Photography)



PentaxForums.com server and development costs are user-supported. You can help cover these costs by donating or purchasing one of our Pentax eBooks. Or, buy your photo gear from our affiliates, Adorama, B&H Photo, KEH, or Topaz Labs, and get FREE Marketplace access - click here to see how! Trusted Pentax retailers:

[dgaies]

I think the shot maybe about 1/2 stop overexposed. When I took the shot I originally had it at +0.3EV and noticed that the playback/review histogram had a lot of space on the right. So I bumped it up to 0.7 and then 1.0. At +1.0 EV is where the histogram filled out to the right, and I don't recall clipping any highlights at the time (although I may be wrong about that as it was right after that that I was told to knock it off and put the camera away )

Anyway, even if this shot is somewhere between ISO 25600 and 51200, I still think John did a great job with it and am quite happy with the overall high ISO performance of the K5 thus far. However, I will try and take a more suitable, properly exposed 51200 shot to see how far we can legitimately push this

[Xatnep_CJ]

What you did bumping up exposure is the best strategy to handle high ISO situations. Any underexposure would result in heavy noise and pushing an underexposed image up makes it even worse. So exposing for the highlights and pulling it down a bit is the way to go exposurewise.

I someone would have told me last week that tens-of-thousands-of-ISO-portraits from an APS-C cam are totally usable I would have called him a crazy liar.

REALLY Good job!

[dgaies]

What you did bumping up exposure is the best strategy to handle high ISO situations. Any underexposure would result in heavy noise and pushing an underexposed image up makes it even worse. So exposing for the highlights and pulling it down a bit is the way to go exposurewise.

I someone would have told me last week that tens-of-thousands-of-ISO-portraits from an APS-C cam are totally usable I would have called him a crazy liar.

REALLY Good job!

Thanks

Yeah, I was just wanted to point out that I certainly wasn't trying to bump up the exposure to try and trick anyone with a 51200 shot that was more of a 25600 shot. Like you said, I was trying to expose for the highlights and didn't (at the time) consider what that meant in terms of creating a 51200 example.

But overall I'm just happy that ISO 3200/6400 is pretty clean and that 12800/25600 can be usable under the right circumstances. So even if 51200 turns out not to be very usable, the K5 has (IMO) proved that it has the capacity to turn out great high ISO images.

[Jochen]

OMG! I just want to swear on russian using bad words! Something like

You Tube

You Tube

[Xatnep_CJ]

... but all of us are waiting for JohnBee's tutorial on denoising high ISO shots.

[JohnBee]

... but all of us are waiting for JohnBee's tutorial on denoising high ISO shots.

'All' is quite a strong word to use.

However... if the sample is bogus, then I don't see the sense in pursuing a walkthrough with this particular file(I could use another though).
Having said that, I am hopeful that we can replace this one with something that feels a little more legitimate in terms of exposure.

[falconeye]

But here it is:

K-5, ISO51200, 1/40s, 85mm, ƒ/5.6

Thanks for the image, the raw file download option and the wonderful processing in Topaz Denoise.

It is a nice manifest of what denoising can do.


As a courtesy for fellow forum members (for easier comparison) I attach the standard rendering (i.e., the default profile) from good old LR 2.x, as a 100% crop.

It shows a lot of noise of coursem but for an ISO 51k, it is still good and as JohnBee has shown, still cleans up in PP.

[dgaies]

Thanks for the image, the raw file download option and the wonderful processing in Topaz Denoise.

It is a nice manifest of what denoising can do.

You're welcome

As a courtesy for fellow forum members (for easier comparison) I attach the standard rendering (i.e., the default profile) from good old LR 2.x, as a 100% crop.

It shows a lot of noise of course but for an ISO 51k, it is still good and as JohnBee has shown, still cleans up in PP.

Thanks for the post of the LR2 crop; it certainly put into context how much a little NR/PP can do. And as you said, despite the fairly high amount of noise, there is still a ton of detail (IMO) left for a 26500/51200 shot

[telfish]

'All' is quite a strong word to use.

However... if the sample is bogus, then I don't see the sense in pursuing a walkthrough with this particular file(I could use another though).
Having said that, I am hopeful that we can replace this one with something that feels a little more legitimate in terms of exposure.

John, please do the example even if it's a 25000 shot. The methods you use and the setting you use will be invaluable to myself and I am sure many others.


Terry

[JohnBee]

John, please do the example even if it's a 25000 shot. The methods you use and the setting you use will be invaluable to myself and I am sure many others.

I understand.
And don't worry.
I still intend on covering this area of the case study.
However, there are very good reasons to cover the highest sensitivities, and I'd like to maintain that if at all possible.
I do have another 51200 sample to work with, but as it turns out, it is underexposed and quite a handful to process.

Having said that, I just got work from dgaies that he is 90% certain to replace the file without an eV push. And so I'm going to continue preparing my workflow(thread) and leave it open for the new sample.

[telfish]

I understand.
And don't worry.
I still intend on covering this area of the case study.
However, there are very good reasons to cover the highest sensitivities, and I'd like to maintain that if at all possible.
I do have another 51200 sample to work with, but as it turns out, it is underexposed and quite a handful to process.

Having said that, I just got work from dgaies that he is 90% certain to replace the file without an eV push. And so I'm going to continue preparing my workflow(thread) and leave it open for the new sample.

Thanks.

I am still trying to work on the original image to get close to your results. I am closer but not really there. Here is a 100% crop from the original (tiff) worked in LR and Denoise 5.