Originally posted by gfmucci I wonder why stills with smaller file sizes couldn't also use proportionately fewer sensor pixels (vs. compression), thus meaningfully reducing heat:noise
gfmucci, you're asking the right questions and just need a little bit more knowledge to become an expert yourself
In the particular case, you were right
IF heat would be the #1 cause of noise and
IF switching off 50% of pixels would significantly reduce heat (which both isn't the case).
Make a test: Compare noise in a shot from a cold camera immediately after power on and after an hour of continued use. The effect is measurable but not large.
For your record: #1 cause of noise is the random part in the photon particle flux from the light source via the subject into the camera lens. Nothing a sensor can influence. Google for "photon shot noise".
In the case of video: The K-7/K-x simply don't have fast enough data pipes to read all pixels 30/24 times a second. So, the majority of pixels are skipped simply. Skipped, not switches off. Video creates a lot of heat...
Originally posted by hcc That is correct if the algorithm to produce 12 Mp JPEG file is the same as the one to produce a 6 Mp JPEG or a 2 Mp JPEG file.
[...]
Has anyone some insight knowledge of the Pentax JPEG programming algorithm(s)?
Be careful. You are possibly confusing two things here. The image is first downsampled (from a preprocessed raw bitmap) and second, compressed (into a file format).
I don't know Pentax' exact downsampling algorithm (it better be bicubic) but it doesn't matter for our noise discussion. The formula I gave does hold true exactly for bilinear resampling.
As for JPG compression: the algorithm is standardized. The only tuning parameter is the JPG quantization coefficient matrix (36 numbers). By selecting an image quality setting (*, **, ***, or ****) you are selecting one out of four matrices. You can inspect a JPG header in a Pentax-created JPG file to recover and inspect the matrix. AFAIK, there is nothing special about it. However, you may wish to inspect it to make sure it doesn't depend on the MP or ISO setting. It may be interesting to note though that smaller files with less pixel noise would compress better.
Summary: your comments are interesting but have no relevance with respect to the question by the OP.