Another reviewer made a great suggestion. Use the Generative models at 1x, then upscale to 2x or more using one of the basic, much quicker models. I've used this method and it works very well.
And it was supposed to be so beautiful! Without a subscription. And here, you have to pay. And everyone returns to the Adobe option> The tutorial itself is as always instructive and brings a lot to the user. In a simple and clear way, all the important changes in this program are explained. Greetings from autumn Poland from the city "Fortress Przemyśl".
I would like to see a hybrid comparison where the cloud is used to recover it to x1 or x2 and then Photo AI is used to upscale it to x4. vs the cloud doing it x4 to see if there are major differences in quality. It could be a way to minimize cloud usage and get the size and quality you want.
Thank you Dave . I have used the functionality and have written directly to Topaz with some queries . They have not yet replied . I have a bunch of scanned images of old family photos which this functionality should be ideal for …recover . So far it has done a good job on faces but not clothes . I have an M1 iMac 16GB ..so used their cloud rendering . You showed the pricing in your YT ..have you noticed that it contradicts the actual pricing . The pricing model per month shows credit by mb size . I seem to have been overcharged in my cloud rendered images because of the anomaly . The charge applied is not that showed below the monthly prices. The refine model is a lot of guess work and yet you are charged for every time you change your mind where slider should be . I think this recover and refine are very much in their infancy .
Thanks, Dave. Very helpful tutorial. Using recovery via the PC takes ages. So not really useful. But maybe they are willing to improve on this. You never know. How it is possible to render so quickly in the cloud, compared to any modern PC … I don’t understand.
Thank you for this informative tutorial Dave. Are the credits that you get cumulative? ie If you don't use any credit in a given month, do they add to the ones you get the next month? And, when you buy credits, do they have an "expiration date". Thank you!
@@spydergs07 Of course, but I have a hard time to see the logic in why Topaz continues Gigapixel AI parallel with Photo AI if it's not to sell redundancy including cloud access with free access to our images. I watched a video from Topaz yesterday promoting Redefine as if it was completed and without endless processing time. The good old Topaz has become a money machine based on long lasting expectations with very slow progress of real advantages. The usable part of their noise-sharpening features are still behind DxO in quality and speed.
What do you mean? You can’t manually add models. Thus far only the provided models they use. (It’s not like stable diffusion where you can use your own diffusion models)
The biggest reason rendering time is that long is not your Intel CPU , it’s the AMD Radeon GPU. The company told me they are incompatible with their AI functionality.
Dave, if you use the gig 8.1 as a plugin for PS , it doesn't show the credits at the top for cloud rendering. However in the standalone version it does. Is this just me or have you noticed this?
Yeah, these guys never let a spoonful of food pass their mouth. I just don't understand why this circus with the program and its download? In my opinion, it was the marketers who, as always, gave the order so that no one could use it for free. They should have immediately posted the terms of purchase for each photo processing on the site. The computer has been working for over an hour, there is no progress in processing, the processing line is at 2x magnification and the processing is not moving at all.
@albertzhager7852 Oh ok, do you have an additional GPU? Any proper AI software is heavily dependent on GPU processing power. Even my 20 core CPU would get bogged down with AI processes. My GPU handles then with ease. I only have a 4060 as well and my images run about the speed of their cloud rendering. When you save an image that is AI edited, your GPU then has to render that entire image out as it sells. When I save an image, my GPU status around 98% the entire save.
@@spydergs07 It's easier for me to use paid services and cloud processing than to buy additional equipment, especially since the program adds all sorts of artifacts that are very difficult to deal with. And in general, I don't have many photos that require this level of processing. I must say that for my needs, Gigapixel 8 does a pretty good job even without AI. I recently restored a photo of my wife from 50 years ago, and the AI drew Sophia Loren's eyes on her. Of course, Sophia Loren is a little prettier, but not by much.
@@spydergs07 "I only have a 4060 as well and my images run about the speed of their cloud rendering." - I can barely believe this. Please do the following test: Take an TIF-image with max. 2000 px at the larger side, scale it to 12000 (!) px and use "Redefine (beta)", creativity slider to "4", export it as a JPG file. For my purpose the 12000 px is still too low. I need at least 18000 px, but 12000 px is currently the maximum resolution you can do with the generative models. My MacBook Pro M1 Max (32 GB RAM) could render the medium sized previews in under 10 minutes. It is currently rendering the entire exported job since over 80 hours now. This does not fit together in my option. I can't say if Gigapixel AI has crashed or if it's still running. A few hours ago I started the same rendering with the same image on my Windows PC with a Nvidia 4070. The previews were rendered in about 2 or 3 minutes on the windows PC, so - as expected - the Nvidia-GPU is definitely faster. It is currently rendering the entire exported job since about 6 hours. Progress bars don't move on both computers. I don't know if both jobs will ever be finished, or if Gigapixel AI (v8.0.2) has (still) a bug, because the "Redefine" model is beta? Or is it the TIFF-to-JPG-conversion during export? Or the "render on your own machine"-option is just a fake?
Remember ANY AI generative program or process is going to be HEAVY on GPU resources. It's just the nature of AI. If you don't have a GPU, this will take you forever. If you buy a GPU, buy at least a 40xx series Nvidia. I have Win11, i7 14th Gen Intel, 4070 GPU/ 8gb ram on GPU, 64gb ram and the Recover only takes 20 seconds Exporting a recovered image is taking me about 3.5 minutes. The cloud option said it would take 2-2.5 minutes. Running it locally, I didn't have to use any credits and it took roughly 1 minute longer then cloud. A Mac with an Intel cpu is fairly old and the M series chips are quite a bit faster, but you still need a GPU to speed things up.
Photoshops' AI is good, but not for scaling and enhancing. I use Ps for extending and some AI work, but to scale an image up, this is a whole different purpose than Ps AI.
Another reviewer made a great suggestion. Use the Generative models at 1x, then upscale to 2x or more using one of the basic, much quicker models. I've used this method and it works very well.
Another very helpful tutorial. Just the facts, not a sales presentation. Bonus Dave Kelly’s expert advice! Thanks Dave!
And it was supposed to be so beautiful! Without a subscription. And here, you have to pay. And everyone returns to the Adobe option> The tutorial itself is as always instructive and brings a lot to the user. In a simple and clear way, all the important changes in this program are explained. Greetings from autumn Poland from the city "Fortress Przemyśl".
I would like to see a hybrid comparison where the cloud is used to recover it to x1 or x2 and then Photo AI is used to upscale it to x4. vs the cloud doing it x4 to see if there are major differences in quality. It could be a way to minimize cloud usage and get the size and quality you want.
Thank you Dave . I have used the functionality and have written directly to Topaz with some queries . They have not yet replied . I have a bunch of scanned images of old family photos which this functionality should be ideal for …recover . So far it has done a good job on faces but not clothes . I have an M1 iMac 16GB ..so used their cloud rendering . You showed the pricing in your YT ..have you noticed that it contradicts the actual pricing . The pricing model per month shows credit by mb size . I seem to have been overcharged in my cloud rendered images because of the anomaly . The charge applied is not that showed below the monthly prices. The refine model is a lot of guess work and yet you are charged for every time you change your mind where slider should be . I think this recover and refine are very much in their infancy .
Well done...great explanation...Thank you.
Thanks, Dave. Very helpful tutorial. Using recovery via the PC takes ages. So not really useful. But maybe they are willing to improve on this. You never know. How it is possible to render so quickly in the cloud, compared to any modern PC … I don’t understand.
Mine runs pretty quick.
I have an i7 14th Gen Intel, 4060 GPU with 8gb ram, 64gb ram and the Recover only takes 20 seconds.
Slow and expensive but Topaz is still the best
Thank you for this informative tutorial Dave. Are the credits that you get cumulative? ie If you don't use any credit in a given month, do they add to the ones you get the next month? And, when you buy credits, do they have an "expiration date". Thank you!
Here is a link to Topaz Labs Cloud Rendering: docs.topazlabs.com/gigapixel-ai/cloud-rendering
This info should answer your questions.
Forcing more processes to the cloud will give Topaz more relevant material for their AI-learning - for free.
Just get a good GPU and do everything locally on your machine.
@@spydergs07 Of course, but I have a hard time to see the logic in why Topaz continues Gigapixel AI parallel with Photo AI if it's not to sell redundancy including cloud access with free access to our images. I watched a video from Topaz yesterday promoting Redefine as if it was completed and without endless processing time. The good old Topaz has become a money machine based on long lasting expectations with very slow progress of real advantages. The usable part of their noise-sharpening features are still behind DxO in quality and speed.
i have installed models for topaz gigapixel ai but i dont know where to put them. where do i put them? thank you
What do you mean? You can’t manually add models. Thus far only the provided models they use. (It’s not like stable diffusion where you can use your own diffusion models)
The biggest reason rendering time is that long is not your Intel CPU , it’s the AMD Radeon GPU. The company told me they are incompatible with their AI functionality.
You may speed up the software by checking if there is a new graphic card driver available for your card? :)
Dave, if you use the gig 8.1 as a plugin for PS , it doesn't show the credits at the top for cloud rendering. However in the standalone version it does. Is this just me or have you noticed this?
I believe that is the way it works. I believe you only get credit options when using Gigapixel as a stand alone app.
@@thejoyofeditingwithdavekelly Thanx, had to be. Appreciate your instructive videos.👍
iMac with intel processor? That's ancient :o
Time to get an M2Max or M4 Pro mini
I just got a Mac Studio M2 Ultra.
Yeah, these guys never let a spoonful of food pass their mouth. I just don't understand why this circus with the program and its download? In my opinion, it was the marketers who, as always, gave the order so that no one could use it for free. They should have immediately posted the terms of purchase for each photo processing on the site. The computer has been working for over an hour, there is no progress in processing, the processing line is at 2x magnification and the processing is not moving at all.
What specs do you have on your PC?
This is a VERY heavy GPU program.
Any AI generative program will be.
This is built for use of Nvidia GPU's.
@@spydergs07 I have a MacBook Pro Cip Apple M1 Max Memory 64 GB macOS 15.0.1
@albertzhager7852 Oh ok, do you have an additional GPU?
Any proper AI software is heavily dependent on GPU processing power.
Even my 20 core CPU would get bogged down with AI processes.
My GPU handles then with ease.
I only have a 4060 as well and my images run about the speed of their cloud rendering.
When you save an image that is AI edited, your GPU then has to render that entire image out as it sells.
When I save an image, my GPU status around 98% the entire save.
@@spydergs07 It's easier for me to use paid services and cloud processing than to buy additional equipment, especially since the program adds all sorts of artifacts that are very difficult to deal with. And in general, I don't have many photos that require this level of processing. I must say that for my needs, Gigapixel 8 does a pretty good job even without AI. I recently restored a photo of my wife from 50 years ago, and the AI drew Sophia Loren's eyes on her. Of course, Sophia Loren is a little prettier, but not by much.
@@spydergs07 "I only have a 4060 as well and my images run about the speed of their cloud rendering." - I can barely believe this. Please do the following test: Take an TIF-image with max. 2000 px at the larger side, scale it to 12000 (!) px and use "Redefine (beta)", creativity slider to "4", export it as a JPG file. For my purpose the 12000 px is still too low. I need at least 18000 px, but 12000 px is currently the maximum resolution you can do with the generative models.
My MacBook Pro M1 Max (32 GB RAM) could render the medium sized previews in under 10 minutes. It is currently rendering the entire exported job since over 80 hours now. This does not fit together in my option. I can't say if Gigapixel AI has crashed or if it's still running. A few hours ago I started the same rendering with the same image on my Windows PC with a Nvidia 4070. The previews were rendered in about 2 or 3 minutes on the windows PC, so - as expected - the Nvidia-GPU is definitely faster. It is currently rendering the entire exported job since about 6 hours. Progress bars don't move on both computers. I don't know if both jobs will ever be finished, or if Gigapixel AI (v8.0.2) has (still) a bug, because the "Redefine" model is beta? Or is it the TIFF-to-JPG-conversion during export? Or the "render on your own machine"-option is just a fake?
I have this program, but as soon as I click on a face the program quits.
What specs do you have on your PC?
Remember ANY AI generative program or process is going to be HEAVY on GPU resources.
It's just the nature of AI. If you don't have a GPU, this will take you forever.
If you buy a GPU, buy at least a 40xx series Nvidia.
I have Win11, i7 14th Gen Intel, 4070 GPU/ 8gb ram on GPU, 64gb ram and the Recover only takes 20 seconds
Exporting a recovered image is taking me about 3.5 minutes. The cloud option said it would take 2-2.5 minutes.
Running it locally, I didn't have to use any credits and it took roughly 1 minute longer then cloud.
A Mac with an Intel cpu is fairly old and the M series chips are quite a bit faster, but you still need a GPU to speed things up.
Cloud? Credits? Plan? POS Not for me.
Never in a million year I'll pay a subscription. 👎
any ai tool that doesnt have generative fill and expand funktionality is useless
pay and pay again! I A is already included in photoshop, so it's not necessary with topaz.
Photoshops' AI is good, but not for scaling and enhancing.
I use Ps for extending and some AI work, but to scale an image up, this is a whole different purpose than Ps AI.