This week’s update contains several tweaks and fixes around the app and a change to the installers. The installers will now download model files for all available devices in your computer and not just the one that would be auto selected. This should resolve issues where people with multiple devices they might want to use wouldn’t have the models available.
As always the change log is 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.
Released August 10th, 2023
Changes since 1.5.0:
- Selecting cancel after turning on Preserve Text will now disable the filter
- Right clicking dialogs should no longer close them
- Renamed “Clear Strokes” button in subject refinement to “Reset Mask”
- Text field preferences now save correctly when clicking “Save” while focused
- Parameters on non-Autopilot selected filters should no longer reset automatically
- Fixed Remove Noise Autopilot indicator sometimes being wrong
- Added tooltips to the Autopilot dot and reset button
- Shift shortcuts are now disabled when inside an edit mode
- Hovering over crop button now highlights image in navigator
- Installer now downloads model files for all available devices
- Subject Only toggle is now disabled when there is no subject
- Change “Edit Subject” to “Select Subject” when subject detection is None
- General preferences reset button should now properly grey out when at defaults
Just a quick question: when will it be possible to use a selection brush to make the face recovery filter more precise? Because sometimes it doesn’t pick up everything. Especially if the hair is long and thick. In this case, some of the hair is in the auto selection but the rest is blurred because it’s not taken into account.
Also, the High Fidelity filter still has the same problems as all the other versions. Artefact problems in most cases. As well as some problems adding a bit of bluish colour in places. Mainly on the edges.
In this image, you can clearly see that there is a subtle bluish colour in the treated area.
The standard filter looks better 90-95% of the time in my photos. The details are more precise.
An example of a result where the Hight Fidelity" model generates blue on non-blue things.
Comparison with the standard model
The bluish effect is still there, but much weaker.
I wish it was possible to upscale using photoshop plugin too like how it is possible in gigapixel.
At the moment Face Recovery is trained to recover the face and some of the hair. Extending the face box further is currently unsupported and could cause a bad output.
It would be best to deselect Hair in this case so the output looks consistent.
Can you upload these two files to our Dropbox so we can take a look and try to reproduce the issue?
We are currently working on this feature! It will be released in the next month so you can upscale with Topaz Photo AI in Photoshop.
Yes, I can send you the photos. I’ll give you the raw file. You’ll have to do some basic retouching in Lightroom classic, then send it in Tif format. Because I think if you send the raw file directly to Photo AI, I’m not sure the problem will be the same.
I am still having problems with whitebalance when I use PhotoAI on my Nikon z8 images.
I have uploaded to your dropbox:
- a video showing you how I am getting the error. “Recording_WhiteBalance_Z8Nef.mp4”
- the NEF-file I am using in my video. “DSC_3358.NEF”
- a dng-file generated by CaptureOne, that seems to replicate the exact same problem when that file is open directly in Photo AI. “DSC_3358.DNG”
I am using the latest available version of Capture One and I’m running windows 11.
Maybe the problem is in how Capture one produces the DNG, and not Photo AI at all, but I can’t tell which.
I did report this in the 1.5.0 version, but got very little response.
That’s it. Photos are sent to Dropbox.
What if you converted your photo to Tif format? Is the result the same?
I’m with you!!!
It is on the Roadmap, but I wish the priority would be accelerated to higher in the punch list for nearer term delivery.
If I export the image as a tif-file it works just fine, but I don’t want to move away from the raw format while using Photo AI.
I wish to keep processing the image after Photo AI has done its magic.
Will it also work in Affinity Photo?
Photo ai Crashing, finally figured out the problem. The only way I could get this to stop was to roll back video card firmware about 7 months, please fix this issue can’t use newest NVidia drivers,
Processor Intel(R) Xeon(R) Silver 4214R CPU @ 2.40GHz 2.39 GHz (2 processors)
Installed RAM 320 GB
System type 64-bit operating system, x64-based processor win 10
Testing the new v1.5.1 on an M2 Mac Mini using 15-year-old quality-challenged non-DLSR digital photos, first with Preserve Text (malted milk container) and general upscaling/sharpening of the 2008 hand drawn signs.
It’s like going back in time and using a better camera!
For those who need it (new purchase, upgrade or renewal), Photo AI is on sale ($40 off) for the next week.
OK, but there should be a support email somewhere on the Topaz site. Maybe the devs here can steer you to the right place, and also ask Support to make the task easier
I haven’t checked, but does someone like Dave Kelly offer an affiliate discount code to let you re-up for less than the usual (rack rate) price?
Here’s a link to his YouTube site (he posts his email there too) where he posts videos pretty much every week discussing/demo’ing the latest Topaz PAI update features:
At the upper (& lower) right of the Topaz Labs site there are Support links to communicate with them. They’re always very nice & reply quickly (ie, usually w/in 24 hrs):