I installed the plugin on a mac in the known folder .../media core - but Premiere 2020 doesn´t even load the plugin to the effects panel and AE 2020 does load it intoithe effects, but can´t load it (Can´t find in-point of effect).
This is crashing AE and throwing up a low level exception on two on both of my Mac's as well (MacBook and Mac Pro). Opened a support ticket and trying different solutions but it's not working on Mac at the moment.
Not working for me either on Mac - followed your instructions (put the folder vulkan into /usr/local/share/) etc - crashes every time when adding the plugin.
I downloaded the trial and noticed in the video image some vertical position jumping while the video was playing. Is there a way to control that so it doesn't happen. Other than that the image looked very good.
I installed the plugin on a mac in the known folder .../media core - but Premiere 2020 doesn´t even load the plugin to the effects panel and AE 2020 does load it intoithe effects, but can´t load it (Can´t find in-point of effect).
Could you please help me out?
Best regards, HB
I have the same issue. Can someone please advise? The plugin is not displaying in the Effects panel in Premiere Pro 2020 on Mac. It's also throwing up multiple errors in AE 2020 too. It's disappointing that others have experienced the same issue before me and yet, there's been no response from the developer.
I am very sorry for the issue for Mac users, sorry for any inconveniences caused.
We fixed this issue, please check the lastest version. Please note the plugin needs the Vulkan GPU library to run, currently this needs to be manually installed. It is simple: download the patch (there is also a copy in ScaleUp_v1.0.1/Mac) and Copy the "vulkan" folder to /usr/local/share/.
Sorry again for this issue, and appreciate for your understanding.
I downloaded the trial and noticed in the video image some vertical position jumping while the video was playing. Is there a way to control that so it doesn't happen. Other than that the image looked very good.
Thank you for feedback.
I think it may be due to challenge footage, and we may not handle the temporal consistency well. Could you open a support ticket, and then I can look into something, if you do not mind. Thanks.
What is the real difference between this and Topaz Video Enhancer?
Thank you,
From my testing, this plugin can produce better details in most of cases with faster processing. In technique, the deep neural network structure and neural engine are different.
Crashes After Effects (PC) instantly. Not a good start for this plugin.
Sorry for this issue.
I suppose you are on MacOS, the plugin needs the Vulkan GPU library to run. Could you please try this: download the patch (there is also a copy in ScaleUp_v1.0.1/Mac) and Copy the "vulkan" folder to /usr/local/share/.
If it does not work for you, please open a support ticket.
Why does the plugin use the GPU resources in varying intervals (between 10% and 100%)?
What are the differences between using the plugin with PP or AE (except the missing context menu option)?
Better / Different Quality? Better Performance?
How do I have to understand the AE-specific option "Render Area"?
Thanks for your interests.
Multi-GPU is not supported currently. Yes, it is planned.
The GPU usage depends on the CPU and GPU schedule, not only for the plugin itself but also the host applications. And the other running applications will affect the GPU usage also. I believe we still have some spaces to improve the GPU utilization.
We do not differentiate the PP and AE, it is expected to have the same quality and performance.
"Render Area" is designed for better performance, it will only render the visible composition area when it is enabled. For example, For a 2K footage, the composition is 4K, and the scaling ratio is 400%, then only part of frame area of the footage need to process. However, if the user wants to do some transform operations after ScaleUp, there will be some black part in the composition area, in this case it is good to disable "Render Area".
I am on what was once termed a "gaming" laptop, a PC, Acer predator 15 inch with 16gb ram quad i7, and an nvidia geforce 980 with 4gb vram. This thing has to get an extra .5 v of power from the battery system to run at full power (not that it drains it, but it really needs the boost; It's also a protection from surges though, small but enough to save from a few shorts).
I used this in a 2020 version of AE. I used it on a piece of old footage I'd already done a blow up for. I had to turn off even the sharpening effects I had put on to get the best image out of this plug, because it had already done the job! Yay! Finally a plug that does some short machine learning before outputting. This footage is some 28 years old from a horrid consumer vhs recorder, and had an interlace. I took care of the interlace with a simplistic plug that accounted for the basic shift, and it was great. After running even that plug in advanced mode, the render was longer, by about 10%%, but it looked sharp, even with the lowest amount of adjusting from the deinterlacer. I tried several configurations on a short section and previewed about 5s. Overall, this plug didn't just do the job, it blasted what I already tried out of the water. I gave up on this before because the $200-400 suite I trialed couldn't do the job. Today, I'm not sure what they would use. I did get their deinterlacer, as that seemed to work great, but I dropped the rest. This product does better, and applies some sharpening to really maintain the details, however, I found that blowing up old footage after de-interlacing works better with less sharpening by other plugs, then this a light touch with this one with only 10-20 followed by a slight unsharp masking, %10 to %20 max with the basic AE plug, but the footage is so old, I may want to keep a slight unsharp look by dropping the unsharp mask (UM). I'll be trying a contrast adjustment later, to see if I cannot bring out the lighting a bit better (really bad lighting and horrible color from this camera) instead of the UM. I was able to correct the color for the most part, and really easy with the dropper, as there is an almost pure white in the room in at least one frame, and a black in another, but it is still a very hazed output. Contrast is what its missing I think. I tried it before, but with the smaller resolution footage it really overwashes it. I'll try now after the blow up and see if it works any better.
I am on what was once termed a "gaming" laptop, a PC, Acer predator 15 inch with 16gb ram quad i7, and an nvidia geforce 980 with 4gb vram. This thing has to get an extra .5 v of power from the battery system to run at full power (not that it drains it, but it really needs the boost; It's also a protection from surges though, small but enough to save from a few shorts).
I used this in a 2020 version of AE. I used it on a piece of old footage I'd already done a blow up for. I had to turn off even the sharpening effects I had put on to get the best image out of this plug, because it had already done the job! Yay! Finally a plug that does some short machine learning before outputting. This footage is some 28 years old from a horrid consumer vhs recorder, and had an interlace. I took care of the interlace with a simplistic plug that accounted for the basic shift, and it was great. After running even that plug in advanced mode, the render was longer, by about 10%%, but it looked sharp, even with the lowest amount of adjusting from the deinterlacer. I tried several configurations on a short section and previewed about 5s. Overall, this plug didn't just do the job, it blasted what I already tried out of the water. I gave up on this before because the $200-400 suite I trialed couldn't do the job. Today, I'm not sure what they would use. I did get their deinterlacer, as that seemed to work great, but I dropped the rest. This product does better, and applies some sharpening to really maintain the details, however, I found that blowing up old footage after de-interlacing works better with less sharpening by other plugs, then this a light touch with this one with only 10-20 followed by a slight unsharp masking, %10 to %20 max with the basic AE plug, but the footage is so old, I may want to keep a slight unsharp look by dropping the unsharp mask (UM). I'll be trying a contrast adjustment later, to see if I cannot bring out the lighting a bit better (really bad lighting and horrible color from this camera) instead of the UM. I was able to correct the color for the most part, and really easy with the dropper, as there is an almost pure white in the room in at least one frame, and a black in another, but it is still a very hazed output. Contrast is what its missing I think. I tried it before, but with the smaller resolution footage it really overwashes it. I'll try now after the blow up and see if it works any better.
Thank you very much for your great testing and feedback.
Hi... Trying out this plug-in looks promising. Do you have a specific recommendation for a gpu that works best with it? I might have a large scale (hehe) project that would make worth investing in the most fit gpu for this plug-in.
Hi... Trying out this plug-in looks promising. Do you have a specific recommendation for a gpu that works best with it? I might have a large scale (hehe) project that would make worth investing in the most fit gpu for this plug-in.
Thanks...
Thanks for your interests.
The source footage resolution affects the ScaleUp processing speed most. If you want to enlarge the source footage with 2K resolution, 4GB of GPU memory is an adequate minimum, however for upscaling from 4K to 8K, you may need 8G memory. And the performance also depends on other factors, such as the number of GPU cores, the GPU memory bandwidth, and CPU performance. There is a benchmark for the popular GPU cards. Generally, the higher the score, the better performance for the ScaleUp plugin. Consider the balance between the performance and budget, the cards such as GTX 1080 and RTX 2070, are good choices, and are sufficient for most of tasks.
Please note, this plugin does not support multi-GPU yet, it will pick one for processing, multiple GPU cards will not help to improve performance much.
Demo working well, but how do I compose the scale/zoom? At present it's just zooming into the centre of frame. What if I want to enlarge one corner?
Also what is the point of the mask?
I think you need more than the 3 page manual.
Thanks a lot for your testing.
For enlarging on the non-center position, there is no controller in the plugin, however you can try to add a build-in transform plugin before it, and adjust the "Anchor Point" or "Position" slider. Alternatively you can also add the transform plugin after ScaleUp, for this case, the "Render area" should be disabled to render out the full composition area for AE.
Thank you for your suggestions, we will update the manual later.
We did not test all the non-centercases yet, if you have issues/comments or any requests, please open a support ticket with more details.
I just purchased the plug-in. I'm running AE version 17.5.1 (Build 47). OS is OS X 10.15.7 (19H15).
Computer is a MacBook Pro (16-inch, 2019) with AMD Radeon Pro 5500M 8 GB.
I’m trying to scale up a video from 854x480 to fit height of 1080 comp.
I’m getting less than 1 frame per second.. so how is this GPU accelerated? My project settings are using metal. I don’t see improvement in fast draft. It seems to be nice quality and it just took 1.5 hrs for a 2 min video.
Any suggestions? I did see the benchmarks and realize I’m at the bottom of the list basically.. but still would love some info
I just purchased the plug-in. I'm running AE version 17.5.1 (Build 47). OS is OS X 10.15.7 (19H15).
Computer is a MacBook Pro (16-inch, 2019) with AMD Radeon Pro 5500M 8 GB.
I’m trying to scale up a video from 854x480 to fit height of 1080 comp.
I’m getting less than 1 frame per second.. so how is this GPU accelerated? My project settings are using metal. I don’t see improvement in fast draft. It seems to be nice quality and it just took 1.5 hrs for a 2 min video.
Any suggestions? I did see the benchmarks and realize I’m at the bottom of the list basically.. but still would love some info
Usually, AI video processing is slow even with GPU acceleration because there is a huge amount of parameters in the neural network structure. It is somehow expected to take around one second per frame for your GPU card. We are investigating to improve the performance without losing quality, and will update it once we make some progresses. Thanks.
I just purchased the plug-in. I'm running AE version 17.5.1 (Build 47). OS is OS X 10.15.7 (19H15).
Computer is a MacBook Pro (16-inch, 2019) with AMD Radeon Pro 5500M 8 GB.
I’m trying to scale up a video from 854x480 to fit height of 1080 comp.
I’m getting less than 1 frame per second.. so how is this GPU accelerated? My project settings are using metal. I don’t see improvement in fast draft. It seems to be nice quality and it just took 1.5 hrs for a 2 min video.
Any suggestions? I did see the benchmarks and realize I’m at the bottom of the list basically.. but still would love some info
Usually, AI video processing is slow even with GPU acceleration because there is a huge amount of parameters in the neural network structure. It is somehow expected to take around one second per frame for your GPU card. We are investigating to improve the performance without losing quality, and will update it once we make some progresses. Thanks.
I really appreciate this plugin - the client liked the results I got and it only took about 1.5 hours to output. Since there are other people asking about the M1 chip, and based on your description, since the M1 has dedicated ML cores, does this mean that possibly one day we'll see slightly faster performance. Don't get me wrong, what you've done here is black magic and I am indebted that I could slam a 10mb file through this and get acceptable enough results out of the ProRes render to make the client happy. That, is mind blowing, as the after effects scale + unsharp mask looked.. less than adequate.
I just purchased the plug-in. I'm running AE version 17.5.1 (Build 47). OS is OS X 10.15.7 (19H15).
Computer is a MacBook Pro (16-inch, 2019) with AMD Radeon Pro 5500M 8 GB.
I’m trying to scale up a video from 854x480 to fit height of 1080 comp.
I’m getting less than 1 frame per second.. so how is this GPU accelerated? My project settings are using metal. I don’t see improvement in fast draft. It seems to be nice quality and it just took 1.5 hrs for a 2 min video.
Any suggestions? I did see the benchmarks and realize I’m at the bottom of the list basically.. but still would love some info
Usually, AI video processing is slow even with GPU acceleration because there is a huge amount of parameters in the neural network structure. It is somehow expected to take around one second per frame for your GPU card. We are investigating to improve the performance without losing quality, and will update it once we make some progresses. Thanks.
I really appreciate this plugin - the client liked the results I got and it only took about 1.5 hours to output. Since there are other people asking about the M1 chip, and based on your description, since the M1 has dedicated ML cores, does this mean that possibly one day we'll see slightly faster performance. Don't get me wrong, what you've done here is black magic and I am indebted that I could slam a 10mb file through this and get acceptable enough results out of the ProRes render to make the client happy. That, is mind blowing, as the after effects scale + unsharp mask looked.. less than adequate.
Thank you!
Thank you very much for your supporting, and glad to know it helped you.
M1 chip shows impressive performance and has dedicated ML cores, unfortunately, this plugin does not utilize the ML cores, it will only use the general-purpose graphic processing unit. And its GPU performance is comparable with the GTX1060, perhaps similar with your current GPU card, so it is not expected to get obviously higher performance. Thank you.
This tool is pretty good! Much better than the built-in Preserve Detail Scale effect in After Effects, it's so sharp!
Though I have noticed a bug that I could replicate twice.
This has happened when I used After Effects with Premiere Pro's dynamic
link. Whenever I apply the ScaleUp effect to a layer (image or video)
with a mask layer above it (originally a solid layer) and then export
the video from Premiere Pro it makes the layer with the ScaleUp effect
invisible and doesn't render, even when inside the AE file it is
visible.
This tool is pretty good! Much better than the built-in Preserve Detail Scale effect in After Effects, it's so sharp!
Though I have noticed a bug that I could replicate twice.
This has happened when I used After Effects with Premiere Pro's dynamic
link. Whenever I apply the ScaleUp effect to a layer (image or video)
with a mask layer above it (originally a solid layer) and then export
the video from Premiere Pro it makes the layer with the ScaleUp effect
invisible and doesn't render, even when inside the AE file it is
visible.
This happened with the version 1.0 on Windows 10.
Thank you for your interests and testings.
I can not reproduce the bug. I am not sure following your steps exactly: In AE, apply ScaleUp to a layer (video) and add the mask (Rectangle/Ellipse) on the same layer, the effect shows as expected. Then in PR, add adobe dynamic link by "import After Effects composition", and add the sequence to timeline, the sequence is shown in Premiere Pro, and then export the media to local disk. During the whole procedure, I do not find any problems.
Does it work as expected without the mask layer? Could you please open a support ticket with more details on each step in AE and PR?
This tool is pretty good! Much better than the built-in Preserve Detail Scale effect in After Effects, it's so sharp!
Though I have noticed a bug that I could replicate twice.
This has happened when I used After Effects with Premiere Pro's dynamic
link. Whenever I apply the ScaleUp effect to a layer (image or video)
with a mask layer above it (originally a solid layer) and then export
the video from Premiere Pro it makes the layer with the ScaleUp effect
invisible and doesn't render, even when inside the AE file it is
visible.
This happened with the version 1.0 on Windows 10.
Thank you for your interests and testings.
I can not reproduce the bug. I am not sure following your steps exactly: In AE, apply ScaleUp to a layer (video) and add the mask (Rectangle/Ellipse) on the same layer, the effect shows as expected. Then in PR, add adobe dynamic link by "import After Effects composition", and add the sequence to timeline, the sequence is shown in Premiere Pro, and then export the media to local disk. During the whole procedure, I do not find any problems.
Does it work as expected without the mask layer? Could you please open a support ticket with more details on each step in AE and PR?
Thank you.
I added a mask not in the same layer but while creating a separate shape layer and placing it above the footage/image layer, then setting it as alpha matte. It exports normally in After Effects but in Premiere Pro it makes the footage/image layer not visible.
I also found it happens when the footage/image layer with the ScaleUp effect applied is below a pre-made transition composition (like the ones in Animation Composer). When exporting in Premiere Pro via Dynamic Link it makes the footage/image layer go invisible in the final video.
When I apply the default settings to a layer, I get a green screen. My GPU has only 2GB of RAM. Is that the problem? If not, what is?
I'm on a deadline and I don't have time to waste. I've spent at least an hour trying to make this plug work. I feel like a developer's crash test dummy who just wasted $50.
I agree with another poster; you need more than a 3-page manual that should include troubleshooting.
When I apply the default settings to a layer, I get a green screen. My GPU has only 2GB of RAM. Is that the problem? If not, what is?
I'm on a deadline and I don't have time to waste. I've spent at least an hour trying to make this plug work. I feel like a developer's crash test dummy who just wasted $50.
I agree with another poster; you need more than a 3-page manual that should include troubleshooting.
Sorry for the problem and inconveniences caused. "If the GPU is not supported, ScaleUp will return a green frame" (From manual). It indicates the GPU is not supported. Could you please open a support ticket for this problem with more details (GPU model, OS version, and AE/PR version), or/and refund request. Thank you.
Could you please check if there is a DLL named vulkan-1.dll in C:\Windows\System32, and please update the GPU driver if it is not found.
If the plugin still can not work well after updating GPU driver, could you please open a support ticket, then I can know more about the details.
Thank you.
sudo spctl --master-disable
then hit return and you're good to go!
sudo spctl --master-disable