This week we have a few new options regard Face Recovery. You now have the option to refine or not refine the hair and neck area for each face. Face Recovery should also now better preserve people’s eye colors.
Along with these we’ve also had several UI fixes and changes which are listed in the change log below.
Please give us any feedback or report issues with this release. We’ll be updating TPAI regularly to address those pieces of feedback and issue reports. If there’s a specific image you’d like us to see, you can send it us at this dropbox link.
I’m sending a clock image via Dropbox that worked very well in PAI prior to 1.4.x. Now the denoising and sharpening are stripping out/smoothing beyond recovery key image details. And, adding ‘jaggies’.
In fact, when I 1st used the image for testing, PAI made it look magnificent and almost 3D. Now it looks like a watercolor that someone spilled water on. Perhaps the denoise/sharpen sensitivity needs a tweak. (I always try all models and various slider strength levels + denoise/sharpen off-on…). Win 11 Pro. Standalone & Ps Plugin.
Later: After hunting to figure out which (beta) release produced the excellent processing and detail - including text. I looked thru the various discussions & can’t find the post to determine which rel. it was. But here’s a copy of the PAI post-processed same image with that earlier release. Converted to a monochrome. Look at the signif. differences in the details… including the earlier rel even made something of the Breguet jewelers mark near the 6-7 hours. [In my personal files I found this as having been processed with whatever curr. beta rel. was in progress on Jy. 1].
Also, for images with multiple faces would it be too much to ask for the preferences to be selected differently for different faces? Some images will have one central face where hair and neck recovery are absolutely needed, but it may not be desirable for others in the periphery.
Here’s a better picture of Mr. Myron Fuller Davis about 1893.
1.)
Some Hair-Areas are not detected and are still unsharp…
Could you maybe implement a “Quick Select”-Tool for the FaceDetection to manually select Areas, which are not automatically detected (like in the “Preserving Text”-Tool)?
In Preferences I have Upscale output size set to 2000px Height however when opening images the height in Upscaling varies between 1998px to 2004px. It is correctable by manualing setting it each time but that defeats the purpose of my default setting in Preference.
I tested the latest release version on a low quality, highly pixelated looking screenshot with some symbols and text and once again TPAI surprised me with a perfect looking result.
If you have Upscale in Preferences set to “Enhance small images” or a scale factor and the input files have different dimensions, the outputs will have different sizes
