Sounds like that could be a very handy “Preference” choice item as the product gets fleshed out!
I totally like the new approach, though, the Photo AI beta Natural scaling model was more like “forgiving” towards artificially made lens dispersion effect - the result (2x Natural, no Face Recovery) is sharpened despite sharpening turned off.
It’s a good result, I’ll admit, but at this point, I would postpone using chromatic aberration postprocessing (ArionFX plugin) until after PhotoAI session.
To close the image, try Control-W (I know Command-W works on Mac). Not sure if this works on multiple images.
The version is shown on the subject line on the display area:
As this is a early release things like output to DNG will be add added as the application progresses.
Just out of curiosity, I tried to upscale your image using Gigapixel AI 6.1.0 and it correctly detected and processed the subject’s face. Though the result looks a little bit off…
Face restoration 50%
Face restoration 100%
Thanks I’ll have to try a gigapixel with it and see how it behaves for me
Win 11. AMD processor. Standalone vers.
Autopilot consistently states there is “Severe Noise” in the images I’m running. I honestly don’t even think they’re medium noise - I’d classify them as Low Noise. Latest pic tested attached. A-Pilot also always states the images are “In Focus” - I find I regularly want to add sharpening to the Autopilot settings (maybe it’s my eyes …).
I like that when I switch to manual settings there’s a Green Line on the slider spectrum (I’m assuming that’s where Autopilot would place its setting). That’s nice! I don’t have to retain a number value in my head and try to match it to the slider bar as I experiment with different settings.
Autopilot accurately detected the Objects and Faces in this image.
I tried Side x Side View this time - working @ 200%. NOTE: The squiggly worm artifacts that resulted in the Processed Preview pane at right (in the green grass). That’s not good. Nothing resembles that in the pic at left. See labelled snip attached.
RE: the Background - The program mostly retained the bokeh effect, but one horizontal streak got sharpened noticeably. That’s an area I’d mask out to avoid it getting sharpened when a masking facility is added.
Generally works well for me, but two of four faces (in a portrait of four people) weren’t detected. I think a button to add extra faces would be useful. The sharpen seems very agressive compared to Sharpen AI. I set the strength to 1 and it was still too harsh. I also miss the mask option so I can choose what gets sharpened. I really like having all three tools in one software package!
It sounds like you’re saying that the sharpening goes up exponentially fast when you manually adjust the slider. Versus gradually across the full span of the slider. Is that right? If so, I see the same thing.
Or, do you think Autopilot over-sharpens? (For me, I usually want to add a bit more sharpening to what Autopilot does, b/c it seems to primarily want to denoise - but does some residual sharpening to compensate for denoise smoothing).
Tried v5 today for the first time. Ran it under Lightroom Classic with no issues so far. Impressed with the speed of processing. Result impressive so far and looking forward to testing further. Definitely a process I will utilize in my workflow going forward.
I’m trying out the “Edit with…” feature of C1 22 Pro build 126.96.36.199 on WIN 10 AMD but I noticed some other strange behaviour. All goes well, picture is delivered to Photo AI, can be edited and seems to be saved. However the TIFF file in C1 is untouched, the changes made in Photo AI are not there.
I managed to get the altered image in C1 by opening the C1 capture folder where the image is stored and deleting or renaming the TIFF file just before Photo AI saves the image and returns it to C1. So it looks like Photo AI doesn’t overwrite the TIFF file in the C1 folders.
I’ve got the same feeling as Fotomaker, but adding additional sharpening is way too aggressive compared to the stand alone Sharpening AI program.
Thanks for the reply. So…does this mean that when you release the non beta software, resize (gigapixel) will not be available in the photoshop plugin? This would mean I would be able to use PhotoAI as a Photoshop plugin to reduce noise, sharpen, enhance, but not resize. I would then have to import the file into PhotoAI standalone or use Gigapixel AI to resize the image, or use the standalone only and then import into photoshop for further processing.
You just need to wait and see how the application progresses as it is still pre-release and I am not privy as to what functionality is likely to be there on release.
This is pre-release software and is not fully functional, as the release progresses there will be more functionality added.
I do understand the concept of beta testing! Aren’t we supposed to report all bugs we find?
The problem with that is if the UI makes it clunky to test then I’m far less likely to do so.
Thing like cropping being missing, sure, fair enough - but making the process of opening/closing images and changing settings more time consuming, nope, that puts me right off. We should be able to focus entirely on testing the core functionality, not fighting the UI to do so.
A clunky UI in an Alpha, sure, but by the time it reaches Beta (though not sure what Early Access is supposed to mean here) you expect the UI to be more or less feature complete. The fact your previous software made closing images more twitchy than it should have been is also not encouraging.
Sharpening in Photo AI is quite problematic generating artificial defects in my pictures. I took the same picture and processed it through Photo AI and Sharpen AI. Photo AI only detected low sharpness, didn’t detect a subject so the whole picture was sharpened. But zoom in at the wooden pole under the bird’s feet. At the left side the structure of the wood is completely changed. Also as Photo AI didn’t detect a subject the bokeh was altered. In Sharpen AI I tested without a mask, the result on the wood is nice.
Top Is Sharpen AI, bottom is Photo AI.
(post deleted by author)
I am using this as a stand alone app right now.
I just started to use Photo AI (PAI) yesterday Sat 8/6/22, downloaded v0.4, opened it and then updated to v0.5. I am running this on my late 2015 iMac, Processor - 3.1 GHz Quad-Core Intel Core i, Memory - 16 GB 1867 MHz DDR3. Here is what I am experiencing:
Started PAI, clicked on browse image, tried to load a 30.4 mb .nef file, PAI crashed, I then restarted PAI, this time I dragged and dropped the same photo into the window, the photo loaded, I then edited the photo and everything worked as it should - quick renderings, excellent results. I then shut down PAI.
Today Sunday 8/7/22 I did the same process, Started PAI, clicked on browse image, tried to load a 30.4 mb .nef file, PAI crashed, I then restarted PAI, this time I dragged and dropped the same photo into the window, the photo loaded. I did not do any editing this time
As an after thought, I went to file/open and tried to open the same photo, this works as it should
Hope this helps