Basic Photoshop settings for RAW JPEG conversion.. EX2F?

Petak wrote:

ttbek wrote:

I personally feel like some of your metrics were subjective rather than objective and that 100% doesn't do anyone favors when they're trying to pick out the very small artifacts/flaws in the renderings. I don't think my posting another image will wind up being constructive but rather degenerate into subjective measures of image quality. I can do it later if you insist though.

Which of my "metrics" are subjective?

Pretty much all of them, unless you have a good way of measuring cleanness, smoothness, detail, magenta, sharpening halo around text, and colour separation?  If you don't have a systematic way of doing that where I can do it and you can do it and we get the same numbers then it's totally subjective.

Are you really going to claim that of the two samples I posted the RT one subjectively or objectively better?

How would I know, you didn't even label which one came from RT and which from LR5.  I can tell after downloading them due to the purple fringe you mentioned, turning on the defringing will fix that right up and turning off or turning down the sharpening will remove the sharpening halo.  Cleanness and smoothness... take a look at the tiered noise reduction.  I'm not entirely sure what you mean by color separation, but final color rendering is largely up to the user, it's not like RT has restricted that in any way.  The RT file has better demosaicing.  Look closely at the B in Beer.  What you will see is that the LR5 file turned some horizontal lines into a checkerboard pattern whereas it is rendered correctly by RT.  As far as I'm concerned you're complaining that RT isn't making all these other processing decisions for you in the way that you like, i.e. you're complaining about defaults.  That's akin to if I left the LR output at 60% jpeg and then complained about jpeg artifacts.  Overall, yes, the RT output is better in the area that we actually need for RAW conversion, the rest is up to your taste and could be further altered with RT or with rasterized image editors.

Furthermore, I am posting 100% samples because they do not contain additional artefacts from up-sizing - you are the firs person to claim that up-sizing does not introduce such.

That's because up-sizing as you're using it is upscaling, in which the image is enlarged, the pixels subdivided, and then interpolated to assign values to the new pixels.  That was not done here, they were enlarged only and the pixels were assigned the exact same color, and no, that doesn't result in artifacts of any sort, everything just looks really bad viewed this way because it's pixelated.  This can be demonstrated with a 1 to 4 pixel mapping, it is reversible and there is no rounding at all, you can test it all day long, there are NO artifacts from this.

Even at 100% the LR processed raw file is clearly and objectively

subjectively, you decided you liked it better, I don't see any kind of measurement.

of higher quality. I would be really interested to see if RT would be able to match this even after spending time adjusting the default rendering. I will be even more intrigued if you are able to post samples

RT is already better in your samples, see the above.

which display RT's "absolute quality" (an objective metric

It's neither objective, nor is it a metric, don't put words in my mouth.  It was my opinion, which does change when the evidence disagrees with it.

used by you BTW).