getting the most noise free high iso from k7

Started Mar 21, 2010 | Discussions thread
GordonBGood Veteran Member • Posts: 6,312
Re: ETTR and high ISO..

A Modest Mouse wrote:

Real Engineer wrote:

When I said, "Cool, do you have a source"

Incorrect, good try though (and I sense you are getting upset from your response). What you said was "The K-7 has a fake ISO 3200 and 6400!!!". Then, when i told you that you were indeed incorrect, instead of actually providing a source of info - you said "Cool, do you have a source." See, in the world of adults - where someone makes a claim (as you have) - they provide facts to back it up. You make a claim, then when that obviously false claim was challenged, you got upset.

It was a simple request for a source so I can expand my knowledge base and have a foundation on which to base that knowledge.

From your last reponse - the one where you said the K-7 had a fake ISO 3200, it would appear you were trying to RELAY some knowledge, rather than asking a question to expand it. So, lets break this down, you made a claim, I challenged it, and then instead of providing facts to back up your claim - you got upset.

Modest, sorry to spoil your rant but Real was probably thinking back to a post that I made stating the proof that all of the Pentax cameras have digitally amplified push processing for ISO 3200 and above based on raw histograms.

What you have written above sounds very interesting but it is not something I can use. e.g. Person X says to me while looking at my K7, that only does real ISO1600, the rest are internally generated digital versions.

Rather than bickering, the best proof that digital amplification is used is when there are missing codes every other level in a raw histogram as happens when cameras just amplify with a digital binary left shift. Unfortunately, it isn't quite that easy to recognize with the Pentax DSLR cameras as in addition to left shifting there is also raw Noise Reduction (NR) applied which serves to smudge the code values together to some extent for higher ISO's, but I think that you will see the pattern that generally every other code is missing from the following K-7 ISO 3200 shot raw histogram (it looks big but isn't so big in file size as it is a GIF:

I assure you that if you do the same analysis for a K-7 ISO 1600 image that neither are there the large sub patterns due to the NR but the histogram is much more continuous with only a very few missing codes that is due to the camera scaling the sensors "natural' gain up to the total range of the raw file of zero to 4095 as a 12 bit output level and the pattern is exactly the same for adjusted exposures for all lower ISO's. ISO 6400 has twice as much a jagged raw histogram as this ISO 3200 image with even more almost missing codes as it is digitally amplified twice as much, but with almost the same sub pattern due to approximately the same level of raw NR applied.

The raw image was downloaded from Imaging-Resource here: and the image is a GIF raw histogram that was generated from the source code provided Java program that can be downloaded from here (the DNG Analyzer): after conversion to an uncompressed DNG using the free Adobe DNG Converter application as the DNG Analyzer doesn't work with the compressed DNG's as come from the K-7.

So I say "The K20D, the K-7, the K-m/K2000D, and the K-x all have these digitally amplified (fake) ISO's for ISO 3200 and above" and it is based on raw histograms such as this where it can be shown that there are these indications of missing code levels where there aren't missing to this extent at lower ISO's.

Not that having "fake" ISO's as compared to analog amplified ones really matters at these ISO gain levels as the noise is much more than just a few missing code levels and further processing down the raw conversion chain will tend to fill in the missing levels anyway.

Regards, GordonBGood

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow