Topaz Photo AI v1.0.10

PAI certainly sharpens/crisps up the image a lot relative to the original posted.

But there is a slight aqua shadow along the profile of the girl’s face and some artifacting in the top part of her hair if pushing the sharpening and detail and face recovery to try to act like a camera-captured image. Motion & Lens Blur don’t do better than Std.

At what size are you wanting to use the processed images? I looked at this one at both 400% and 100%…

Pre-Processed Look:

Processed at Diff. Zooms:

1 Like

Photo AI does render out DNG files, which retain the RAW editing capabilities, which is what I am looking for., as is Lightroom Classic. So, a stand-alone Photo AI with proper batch processing, or a working Photo AI plugin for Lightroom Classic would be the solution that I am looking and hoping for.

Maybe I am using it incorrectly, or does anyone know whether Photo AI does allow the scale parameter to be applied to all files in the strip of imported files? Maybe there is a terminal command that harnesses Photo AI’s processing capabilities?

There is an external editor interface for Lightroom… just make sure it is not running when you do a install of Photo AI.

The Command Line interface for Batch
process is maybe not yet complete but the Help is there …

Type: PhotoAI.exe --help and all options will be revealed.

Thanks for the pointer, Don!

I’ll look at that and see whether anything is there to get me closer to where I wanna be.

I see that you’re on Windows, while I am on Mac…haven’t been able to get PAI to reveal its Terminal capabilities to me. When I run “sudo /Applications/Topaz\ Photo\ AI.app --help” from Terminal, I only get “sudo: /Applications/Topaz Photo AI.app: command not found”.

Any ideas?

Not sure, just make sure the address for PhotoAi.app is correct and is there.

I believe it was introduced in v1.0.4 release.

I dragged the app into Terminal, to make sure that the path and free spaces are entered correctly, then added the “–help” parameter, so that part should be correct.

Looks like I’ll have to get in contact with tech support to help me with this.

Thanks for trying, Don!

1 Like

I tried submitting this same bug had in version 1.0.8 on the main site and it got moved to the 1.08 release thread, I get this ‘bug’ on both my systems (win 11 and Win 10) Running latest Photo Ai and Latest Lightroom Classic (12.0.1), Photo Ai opens then nothing.

Relatively often the colors of an image are alienated when using Photo AI 1.0.10 and get a “green cast”.

So it seems in the past couple of releases, Photo AI gets slower and slower. I have an RTX 3090 TI with resizable bar enabled, and I swear this thing runs faster in CPU mode than GPU mode. The UI is almost completely unresponsive. It takes about a minute to process a 512x512 face (from a DeepFaceLab faceset) to 1024x1024 with face recovery and 2x enabled (I used to be able to do about 20 seconds an image with 2x, face recovery, sharpening and noise reduction enabled). GPU Z and Windows Resource Monitor shows the GPU is hardly being hit at all (I think its all from the GUI, I don’t think any processing whatsoever is happening on the GPU)2022-11-24-03-42-45.tzlog (802.2 KB)
2022-11-24-03-46-1.tzlog (1.6 KB)
2022-11-24-04-25-42.tzlog (753.6 KB)
Topaz Photo AI Installer Log.txt (4.3 MB)
2022-11-22-21-36-42.tzlog (2.0 MB)

Thank you very much for helping. When I use PAI to process pictures captured in a video, when using facial restoration model, the image quality of the face and the body is very different.
I am using the same parameters as you, and there is a circle of contour around the face with obvious changes.

When I closed the facial restoration, there were many messy lines on the photos, which looked like oil paintings.

When optimizing pictures captured in a video, it seems that the processing of face and body is not coordinated enough (the first picture below). If facial restoration is turned off, the face will look more cluttered (the second picture below)


I am looking at this picture at 100% and 200% for my screen resolution is not big enough.

Not sure if this has been reported before, but for me, the Strength slider for noise reducion in Topaz Photo AI only works with raw files. With non raw files (e.g. TIFF), this slider does nothing, and apparently, noise reduction is applied at its maximum value (or so I think based on the image appearance). Same goes for the Detail slider.

I guess this must be a rare issue that just a few people, or maybe just me, are experiencing - otherwise there would be more people commenting about it… (?)

The Sliders for the Sharpen section do work, though.

I’m using Windows 10, and at this moment, Topaz Photo AI 1.0.10.

A couple of points regarding the use of “File, plug-in-extras” approach to using Photo AI with RAW files within LR Classic.

  1. If using from within a collection (which is my typical workflow), I’m returned to the original folder that contained the RAW image - not returned to the collection like happens when using the Edit function. Definitely adds steps then to get back… have to move it to the collection and then go back to working in the collection.

  2. Metadata does not carry over. The first time I used it, took me a while to find the photo. I’m bad when I travel internationally about changing my camera clock - so usually just change the photo time via metadata hour(s) shift. Also keywords are lost, etc.

  3. Interesting that the size is different (slightly larger) and I’ve had a few that look like a lens correction was applied while in PhotoAI on the RAW file. Not sure if it was something done in LR or PhotoAI because I don’t have automatic lens correction setup in LR.

I like PhotoAI a lot - still learning about where it fits best. It is masterful at recovering faces in older photos that have been scanned in.

Good work - I’m sure it will continue to improve.

What I meant was go here and open a chat window and email support…

Topaz Support

Just installed the new version and tried it out on a few photos this morning. Worked great. After the celebrations today I need to take a closer look at the settings but so far my only complaint is that I don’t see an option to disable auto preview and after saving the image the image is processed again automatically and then when I close it I’m told that changes have not been saved. It’s kind of annoying to have that after just saving it and then having it process the image a second time after the image save dialogue window has closed. Maybe I’m just missing a setting but I would love to disable auto preview. Also miss having a simple X to click to close the image after processing. Guess I’m just use to the standard interface as Denoise AI and Sharpen AI. None the less, amazing product as always. Happy Thanksgiving.

When using “Process in Photo AI” with Lightroom Classic, Nikon Z72 Nef files saved as DNG from Photo AI 1.0.10 do not retain the lens information that can be used by LR to correct distortion via embedded lens correction information. Not good when processing images from a Nikon Z 14-30mm f/4 :slight_smile:

OS: Windows 11

Feedback…

When taking pictures of a stucco house, the program thinks it’s noise, auto-selects STRONG, and smooths out the actual/normal “grain” of the stucco. Which is not good of course.

Thanks…

You mention seeing similar things to what I was seeing. So it’s not just my PC…

The moat-like shadow outlining the face and “fairy dust” in the hair were most notable on my display. And, while the image got noticeably sharper, it was still soft.

The only way I could get it to where I did was with the Face Recovery. I suspect some of the issues we both found was as a result of how that model/code works… it will be interesting to see what happens as/if a more controllable Masking feature is added that will let us determine exactly where we want sharpening, denoising and face recovery applied iteratively (in other words work with one area we precisely control, then separately another, etc.).

Thx it helped to see your image examples!

This software makes a very good job! It would be good to have all the options available in the autopilot configuration. I would have a lot of photos to process as batch. Now it’s impossible to set “lens blur” and disable “raw noise” for example on all pictures. The only thing I could do now is setting an autoclicker with Python, which is quite stupid and would take ages…

Thanks in advance!

1 Like

Hi there! loving your app. It would be even better if it just copy metadata to the resulting image (like gps). All my photos have gps but, when i edit them on topaz photo ai, the resulting image usually got no gps (other data yes).