Originally posted by Adam In retrospect it would probably be better to stick to in-camera JPEGs for a comparison like this (as long as all settings are the same), as there are so many variables in RAW files these days that you can goof even if you go way out of your way to try to get two files to look the same.
I wouldn't use JPEGs, because then you are also comparing the in-camera JPEG engines which are tweaked all the time.
I'd also recommend a basic converter such as dcraw. RawTherapee may also work, but in any case you need to make sure that exactly the same WB settings and the same tone curves are used.
Originally posted by Adam Aw well- I can take it out again tomorrow! Test charts would also work, of course, but that's not what it's about IMO. Been trying to avoid those as much as possible!
It would be great if you could take them out again.
I'm happy for you to avoid test charts but real world scenes are more challenging to get right, i.e., make sure the focus is spot on and identical for both cameras.
Originally posted by jaad75 Why it should be avoided?
ACR is a very capable converter but it does a lot of things behind the scenes.
Depending on the camera model, it will use different black point settings, tone curves, etc.
Using the same slider values in ACR (and LR) is no guarantee for achieving the same RAW development.
Hence it should be avoided for camera comparisons as it is very hard to eliminate any difference introduced through ACR.
Originally posted by bossa When you remember that the screen res' is 72dpi, and that the K-3 is being downsampled to 16MP, you really have to think that this test was "inconclusive".
Screen resolution is irrelevant when -- as Adam did -- you are allowed to zoom in to a 100% view.
Also, the downsampling to 16MP isn't as crippling as you describe it, as the image will gain with respect to both noise and micro-contrast in the down-sampling process. It is just a bit better to upscale the lower-res image, but it isn't as terrible as you make it appear to create a level playing field by downscaling.