This week we’ve added some new non-RAW Remove Noise models! These have been in beta for a few weeks and will now be in the app alongside the older models for a period of time. These will eventually be replacing the older models, so if you have feedback about the model quality please post in the dedicated model quality thread. We want to try to keep discussion there since both sets of models will be in for multiple versions.
You can find a list of the rest of the changes in the list 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.
We’ve recently added the ability to post RAW files under 65MB in the forum. If you are okay with your image being posted in this thread publicly you are free to do so. You are also free to go through our dropbox if you do not want to post publicly.
Added new v2 non-RAW Remove Noise models
– These will eventually replace the old models.
– If you have feedback please post in the dedicated model quality thread.
Added camera profile for Galaxy Note 8
Improved messaging in various auth states
CLI will now preserve folder structure when using --recursive and --output
Fixed CLI not respecting enhance Autopilot options
Fixed downscaling not working
Fixed exposure on S22 Ultra rear wide camera
Fixed issue where unknown cameras would be prevented from saving
Fixed crop not being correct on some RAW images
Fixed crop issue when XMP was rotated 90 degrees clockwise
Fixed crop issue when XMP crop used a different tag name
Fixed “Could not read image” error messages not having any information
Known issues:
HDR or Monochrome DNGs which previously did not open may have issues when saving as a DNG
– To get around this you can save as JPEG, PNG, or TIFF which shouldn’t have the same color issues
Hi - can you please tell us what face recovery issue you’re running into? For the activation issue, please send us one or two of your log files from 1.3.4 from Help > Open Log Folder. Thank you.
EDIT: FIX FOUND, PROBLEM WITH PC. No problems with the program or concerns of GPU damage!
OLD Post
Upgrading from 1.3.2 to 1.3.4, adding several or many 48-bit 344mb 60mp.tif photos, one loads, then processing starts on it, then my rtx 4090 gpu does loud coil whine then the computer loses power (my computer has no BIOS battery so it resets each time). This happened twice, no overclock on gpu.
I don’t want to risk running this version any more for fear of damaging my GPU
(coil whine isn’t anything to worry about its the computer losing power because of potentially what the GPU is doing concerns me, though stress-testing the GPU it shows no signs of damage in Unigine Heaven Benchmark),
so I won’t be able to test this revision any more. I can try 1.3.2 again to see if the problem persists and is a problem with my system or drivers instead.
Thanks for the hard work by the way everyone, this software has gotten really awesome by the time of 1.3.2 .
face recovery was just sharpening faces instead of using AI recovery, as I said going back to 1.3.3 works so it’s definitely an issue with this newest release, here’s a log from before I rolled back to 1.3.3
Thanks for the new Denoise models. I have tested them on TIFF files and they look much better than the old ones. I can control the strength slider from 0-100 – with a clearly visible effect and I can also keep control over sharpening artifacts. On the whole a very welcome update and step forward. Keep it going!
I downloaded 1.3.4, but I have not yet installed it. The posts above are very concerning. Especially about possibly damaging a GPU. I did not know software could do something like that.
Hi - we haven’t run into this issue while testing on RTX 4090. can you by chance zip all your log files in Help > Open Log Folder and send them to me? I want to check the ones from v1.3.4 to see if anything stands out. If you also know what your power supply is, please let me know.
Hi Radel, for Face detection/recovery detecting areas like fencing…etc, can you send me the image(s)? If Face Recovery isn’t enabled by default on those objects, they must be selectable low-confidence faces (false positive) so I want to verify that. https://www.dropbox.com/request/omTTisNIIfxFDX0eYnyL
For the crash on Strong noise reduction and the Activate issue each time you open the app, can you please recreate both issues and send me your last two logs from Help > Open Log Folder? Thank you.
Here are the log files asap. I don’t want anyone to be concerned because GPU isn’t showing any problems or signs of damage .
Could potentially be PSU because although it’s 1200w it’s 4 years old and the computer was meant for a 2080 Ti pre-built. EVGA 1200W P2. I think these newer cards are meant for PCI Gen-5 PSU’s or something and this one was around for Gen 3.
All the best!
Edit: Just to be sure this is the first version of Topaz Photo AI and first program (run newer games a lot) that’s caused the computer to crash/lose power like this. I definitely can test 1.3.2 again to be sure this isn’t a new problem with my PC . GPU did run at 600W OC’d no problems with NVIDIA Geforce Experience tuning and goes above 500W with PerformanceTest compute tasks, but doesn’t coil whine except in Topaz (not sure how far back in versions). Again, GPU wasn’t OC’d and running with 450W stock limit during these crashes.
I also have a problem with face recovery in version 1.3.4. However, only the preview is affected. When I save the photo the faces are enhanced. I hope the preview problem can be fixed quickly
Tested in 1.3.2 and problem occurred again (albeit when processing images instead of previewing them).
Problem seems to be of all things the new RAM. New RAM doesn’t run at spec’d settings with this program so underclocking fixes it. This program just is intensive enough to make the RAM trigger errors probably.
There wasn’t even any coil wine with the ram down-clocked. Getting 1.3.4 installed to test immediately, will edit post with update of test.
Edit: Problem addressed. No problems in 1.3.4 after testing with downclocked RAM (just to extra clarify, system memory and not GPU vram). Not even any coil wine in the GPU.
Sorry for any concerns I’ve caused anyone , I’m glad to have identified the problem on my end .
I have tried about five images and I don’t see any difference in faces with recovery turned on or off. I don’t think it is working either. All images are using enhanced resolution.
Update: Here is the preview with face recovery on:
.
.
, I’m glad to have identified the problem on my end 
