Tried this rel. earlier. Closed out PAI for several hours.
Tried new images. Still getting same odd bhvr noted in test earlier.
When I close an image I’ve worked on (either using Close Image, Close All Images or CTRL + W) so that no image appears in the UI workspace then I open another (diff.) image, the image I’d just closed before opening the new image flashes up before the new image opens.
BTW, I’ve been using raw images to test so far - a mix of ARW and ORF and raws from those sources converted to dng using Adobe dng Converter.
It’s happening every time in the Standalone. It isn’t a one off experience.
TPAI 1.0.3 may have problems creating output files after having optimized large TIF:s. I repeatedly get “Error creating output file” independent of to which file format I save. Saving to original location, lots of space, lots of memory, fast computer.
V1.03 is crashing randomly on me when processing a stack of images using autopilot. This happens when using folders on local SSD or on remote NAS drives. GPU is an Nvidia RTX 2080
I too am having this problem. Win 10, i7 7th gen, Nvidia 1080, 64 GB ram.
I load up about 200 .nef files and let it go to work, it gets through about 30 or 40 files before it closes without warning. I have plenty of disk space.
I can simply open the program again, drag and drop the files where it left off, and it will start again, so it’s not dying on a bad file.
Would be nice if it would give off some kind of death rattle, any type of error to indicate what’s happening.
Also I noticed a bug where if I load some files up via drag and drop, then tell it to remove them all, I cannot just drag and drop the files into Photo AI again. I can manually tell P AI to open them by navigating to the folder through the interface, and that worked fine. But no more drag-and-drop until I closed and opened the program again.
There is an ascii text log file, I’m running it again right now to hopefully catch something useful. It seemed to not like my display calibration last time, which completely confuses me… I use the Calibrite system, which should be pretty normal as such things go.
Something is wrong with the new Standard sharpening option. When it’s enabled in combination with Face Recovery, the backgrounds - along with the edges of some faces - get offset diagonally. This doesn’t happen with Lens Blur sharpening. The effect occurs on the display and in the exported results. Compare these examples to see the offset. (I’m working with JPGs in standalone mode.)
Just bought this version. So far not very impressed I’m afraid. I’ve immediately hit two serious issues within 20 minutes of using the software.
The first one is where the face refinement not processing the bottom 3rd of the first portrait image I tried. There’s a clear point on the image below which the face is not processed. It’s a full face portrait, so the entire face is selected.
Second problem is the program crashes right out every time when saving a processed Canon CR2 image as JPEG.
I’ve sent logs etc to the support email. Really hope you can sort out these disappointing issues. Feels like I’ve purchased a beta.
A game of two halves !
Firstly the plugin is the most usable version of Photo AI yet !
Sharpening is far better , processing is via GPU when selected, and fast.
Everything seems a notch up , bar a slight colour error in the preview window shown . Tested and the reality is as per Photoshop when processed, so only a slight preview issue .
Now the problem… the standalone seems fine until processing when it crashes . Tried DNG and Tiff saves and crashes after a moment and nothing written to destination. Using files from my test folder that have been through the previous versions fine from CR2.
Still wish the program would follow the Windows 11 screen partitioning standard and allow area selection.
But that’s not the point, for example DXO software gives perfect colors. Obviously I know I can fix the color or even the tonal curve, but the software should be able to output a file with proper color fidelity like it does with other cameras.
Hi! I can’t seem to get batch processing to work. I have large image sets (essentially an object on a turntable with a photo at each degree) – and need to apply the same settings across.
Is this feature working? Or is it possible to implement, either by GUI or command line?
I have been playing a bit more and it is behaving as a standalone with a few of the errant files , but managed to produce a crash when set to “ens corrections” and output to DNG. 2022-10-07-15-34-6.tzlog (11.1 KB) 2022-10-07-15-27-24.tzlog (667.2 KB)
I just discovered that the diagonal offset appears only when using the GPU for processing. There is no problem if I switch to CPU only. And the entire image is offset diagonally if Face Detection is not active.
To summarize … the offset only occurs when Standard sharpening is combined with GPU processing, so something is still ‘wrong’ with the new mode.
Here the log for a new tentative same file , absolutly repetitive, program close without saving when I push save in the save dialog . 2022-10-07-19-56-13.tzlog (92.2 KB)
Nice to have more sliders on Remove Noise, however, we still cannot (pre-)select the AI model for remove noise filter.
So, for my test picture, Autopilot chooses “medium luminance noise” and still the slider “Strength” does not have any visual effect. The noise is just heavily removed, no matter if Strength is 1 or 100.
