Gigapixel v6.1.0

Same problem here.

Version is 6.1.0 (upgrade from previous version with online installer). Missing model files downloaded.

My specs:

  • Intel(R) Core™ i7 CPU 960 @ 3.20GHz, 3201 MHz, 4 cores
  • NVIDIA GeForce GTX 1070
  • 12 GB RAM
  • Windows 10 x64 21H2

the method @Bart_van_der_Wolf mentioned that downloading models when needed only, rather than all of them at once, will help.

If you want all the models now, our support team (there is a chat button at https://support.topazlabs.com/) can help you.

Cool new feature guys. Already having some really great results. I will echo the issue brought up by others in this thread that there seems to be a blurry “square” around heads when using face recovery; more prevalent on higher values. Sometimes its hard to see and sometimes its blatant and photo ruining. Excited to see how this feature is improved and refined going forward.

Just a quick warning, make sure your GPU drivers are up to date from the manufactures website.

For clarity’s sake also use the Help- Graphics info. menu, press the Copy button and paste that info here.

Nice work @alexander.zhang , the face recovery works well, even better than the beta, on a heavily compressed image …

Model files CAN’T be downloaded all at once, but individually. My internet is fine since I downloaded the installer.

6.1.0 is radically slower on my Mac Studio than 6.0.0. I have a folder of benchmark images – 6.0.0 would run through batch processing at ~25 images per minute on average. 6.1.0 is running through 3 to 8 images per minute. It’s definitely not hitting the CPU hard, either…very low usage rates overall.

Anyone noticed a problem with crop in this version? I just tried a few photos and the instant I click crop they get squashed horizontally so its unusable.

Seems to happen on Windows but not Mac.

Is that with face recovery on or off?

I’d expect it to be slower with it on as its far more intensive than the old version that just seemed to dial-down AI on faces rather than actually try to reconstruct them.

It also increases with every face it detects so a crowd can take ages, but its also where it excels as its lots of smaller faces which upscaled poorly before.

hi.
I have installed Gigapixel version 5.8.0 on macOS 10.14.6. When I try to download the new version at the end of the download I get this warning:

“Unable to download component Application: Component file version does not match”

What can I do?

Are we gonna get an update that improves color preservation? There’s a few files that look off when I upscale them. For example an image with lots of purples will get that area look saturated to the max making it look super off (looking like the colors became a very dark purple/blue).

I would share an example but don’t know how.

Download the full installer and install, instead of using the in-app installer.

1 Like

Direct message me.
Check if you can see the same thing on both cpu and gpu processors if the pc has a gpu.

how can I direct message you?

update, I checked the CPU and GPU processors and still see the same results, also just figured out I need trust lvl 2 or whatever that means

…same here. Tried it multiple (at least 20) times, without any success. There is NO problem with my internet connection…

Hi.

I solved the issue with the following steps:

1 - reboot the computer after installation
2 - Doewnload the files without having any file opened in Gigapixel.

This worked for me.

Greetings,
Bernd

1 Like

I resized a batch of photos (70, 2X, Standard) on 6.0 in 1:57

Upgrade to 6.1, same resize in 0:54

Wow! That is significantly faster!

No such luck here. The face recovery bit of GAI still can’t find the model and fails.

18:41:55.318 ERROR [0x704] No usable backend available (:)
18:41:55.318 ERROR [0x704] No backend runner available (:)
18:41:55.318 WARN  [0x704] ERROR!!!! Face parsing model load FAILED (TEUpsamplingWorker::faceParse:1187)
18:41:55.318 INFO  [0x704] Face parsing completely failed. Will return nothing. (TEUpsamplingWorker::procMatAIEngine:267)

I messaged you for full log.

If you installed with in-app updater, try installing with the full installer.