I wanted to ask how to export correctly my processed file as JPG for sharing.
Here is an example of one of the files (from play_raw - [PLAY RAW] Vulcan stone sunset - #99 by Suki2019). I was just playing around and when I exported, the jpg file contains strange grains (image below - Left is as seen in DT and right is after export). Both had the output profile set as sRGB.
I assume you use darktable 3.8.0. Please correct me if this is wrong. What operating system do you use? Is a display profile set up? Is your viewer colour managed? Can your viewer properly scale down large images? Or are both of the screenshots at 1:1 (100%, pixel level) zoom?
In darktable, do you have any options enabled that could badly affect quality / cause a difference between preview and export? Note that not all differences can be eliminated, as the darktable preview window uses some shortcuts to speed up processing. Namely, do you have general → prefer performance over quality or darkroom → reduce resolution of preview image enabled?
Could you perhaps share your raw, your xmp sidecar, and maybe the exported JPEG here?
The DT version is 3.8.0
Operating System - Windows 10
I am not sure about the display profile being set. It is what it is on my Laptop. I have never altered it.
The screenshots are side by side. I was doing best fit on DT and it doesn’t say the zoom level then. I tried to be as close as possible.
The raw file is mentioned in the original post (Its not my file so I dont think I should (re)post it here).
Edit: To my surprise, the rendering here is spot on. Maybe its my viewer (using nomacs)
Tried the same in irfanview, the rendering becomes correct. However when I resize, temporarily the image becomes similar to what viewed in nomacs, but correct automatically.
Some viewers use low-quality scaling algorithms, as they are faster. Others (and Irfanview is probably one of those) render the image twice: first, a low-quality, fast algorithm is used (to give you a preview, improving user experience), then a higher-quality, slow(er) algorithm is applied to eliminate artefacts. That would explain your experience (first you see something similar to the rendering you had issues with, but a moment later the photo is displayed correctly).
Thanks for the clarification. This makes sense for sure.
I have additional questions based on this - do such processing, where the low quality rendering can produce artifacts, be avoided? /if so, any suggestions on how to?
disable general → prefer performance over quality
for the export, if you want to save a downsized image (so you set the output resolution to a concrete value, instead of 0 x 0), enable the high quality resampling feature (the 2nd option under size, in global options of the export module).
In the viewer: check your viewer’s preferences; I don’t really use Windows, so cannot recommend any specific tool.