сморк · 07-Фев-20 20:28(4 года 11 месяцев назад, ред. 01-Сен-20 02:00)
Topaz Video Enhance AI (Topaz Gigapixel AI for Video) Год/Дата Выпуска: 2020 Версия: 1.5.3 Patch notes Разработчик: Topaz Labs Разрядность: 64bit Язык интерфейса: Английский Таблэтка: вылечено Минимальные системные требования:
• Windows 7,8,10 x64
• Только Nvidia
• Nvidia 750Ti (CUDA 3.5) or above with at least 2GB of VRAM.
• Процессор Intel CPU from 2015 onwards (4th generation) или AMD CPU from 2016 onwards
• 8ГБ ОЗУ
• Последняя версия драйвера Nvidia, OpenGL version 3.3 or later Рекомендуемые требования:
• 16 GB RAM (Optimal - 32 GB RAM)
• Nvidia RTX series or above with 6GB or more of VRAM
• Процессор Intel i7 (7th generation) or greater (4GHz and above) или Ryzen 7 or greater (4GHz and above) Описание:
Topaz Video Enhance AI - это переименованный Topaz Gigapixel AI for Video, который увеличивает видео, используя искусственный интеллект. Прямой потомок программы Topaz Gigapixel AI Доп. информация:
Поддерживает видео в форматах: avi, flv, m1v, mkv, mov, mp4, mpg, mpeg, webm, and wmv, а также наборы изображений-кадров в формате png: image_000000.png, image_000001.png...
Вывод либо в mp4 , либо в набор изображений (png,tiff)
Старые видеокарты пока не поддерживаются. Скидка 15% на все продукты Топаза http://topazlabs.refr.cc/topazl
Сброс триала https://topaz-reset-trial.glitch.me/ Скидка 100$ (только для VEAI): проходим по ссылке и вводим купон veai99
Особенности репака
Вылечено. При запросе ввода логина\пароля просто жмём ОК.
Описание моделей
Theia-Detail: EU,P - Output with more details but may have artefacts.
EU: enhance and upscaling jointly.
P: for progressive video only of moderate/high quality. Theia-Fidelity: EU,P - Output with higher fidelity but less details
EU: enhance and upscaling jointly.
P: for progressive video only of moderate/high quality. Gaia-HQ: P,HQ (Previous name HQ-Normal) Upscaling video
P: for progressive video only
HQ: input video is high-quality Gaia-CG: P,CG,HQ (Previous name HQ-CG) Upscaling for computer-generated content
P: for progressive video only
CG: computer generated content
HQ: for high-quality video Artemis-HQ: P,HQ,MC Enhance and upscale video
P: for progressive video only
HQ: for high-quality video with little noise/blockiness
MC: keep motion consistancy (less flickery) Artemis-MQ: P,MQ,MC Enhance and upscale video (удалена в 1.2.1, т.к. результат идентичен Artemis-LQ)
P: for progressive video only
MQ: for video with some noise/blockiness
MC: keep motion consistancy (less flickery) Artemis-LQ: P,LQ,MC Enhance and upscale video
P: for progressive video only
LQ: for low-quality video with strong noise/blockiness
MC: keep motion consistancy (less flickery)
Fixes Mac Crash Fix
Fixed an issue on Macs where Gaia would crash on all machines and all models would crash on pre-10.15 machines Windows Openvino Support
Fixed an issue where Openvino wasn’t properly getting selected Windows missing DLL
Added a DLL for those who have not already installed Microsoft Redistributable Package 2019.
1.5.2 (27/08/20)
A5557B36466A5615A093ACE27458ADC40A76B554 Fixes Option to Select H.264 or H.263 for Video Output
File sizes are smaller with H.263 than H.264, but H.264 is better quality. We consider this a temporary solution while we re-write our video IO engine to better accommodate different formats.
UI Fixes and Messaging Improved
The phrasing of errors and messages has been changed along with some new notifications to give the user more choice in how the application behaves.
1.5.1 (13/08/20)
1633EA24B2FD1EC2D5F3DABD78DB60C02B7F1990 Major features Saving Model Parameters as Default. The last release broke the ability to save the theia model parameters as default. This should fix this issue. Switch to H.264 video output We have switched the codec from H.263 to H.264. 1.5.0 Major features Auto Save/Load. Now as you make changes and process videos we will save your changes so if you exit the program you can click to reload your last save. We will even save progress through an image sequence! Improvements
Theia Model Update
Stability improvements and bug fixes
1.4.2 (30/07/2020)
Хэшкод для торрент-клиента 7D4776A5887A1631E82083B3A4BC59DBF58EAE53 Major features Command Line Interface Revamped the command line interface to make it easier to use and created full documentation Improvements Theia Model Update The Theia model has been updated with a new version.
1.3.8 (09/07/2020)
хэшкод для торрент-клиента: 5B59D07ED9397C112D6F8B33F6238FDD037E2DE6 Change from RC-1.3.7 is minimum:
- Theia-Detail and Theia-Fidelity model update
- New slide at the beginning of of the program In addition to improving usability, we set out the V1.3.x development with 3 key objectives:
- Improve the quality of upsampling, especially to reduce detail flickery problem.
- Increase speed, especially for Mac users.
- Better video IO handling. The new Theia models produce more motion-consistent, therefore more natural-looking upsampled videos, but we are still trying to improve its detail generation. The speed increase is noticeable on both Windows and Mac and will help our users to increase productivities. So objective 1 and 2 are advanced. Unfortunately, we are still working on video IO improvement.
We appreciate very much your helping us both with great suggestions, pointing out problems, and testing. Without your help, there is no way we can achieve this outcome, given the nature of this product. Your help in testing on diverse types of computer systems and different use-cases is absolutely essential!
We are now working on V1.4.x development. We count on your continued support to make it a product that can help you the most. In terms of Gaia models, its structure is weak in dealing with motion consistency (strong for details), while Theia is designed to be better at motion consistency.
1.2.3 (07/06/2020)
Хэшкод для торрент-клиента: F773D05873FE3EC3D2556B4122F870725E6FFAED Improvements
Improved Gaia-HQ model to match the quality in version 1.1.1.
1.2.2 (07/05/2020)
Improvements Improved performance speed for Mac versions. Longer video processing times should see the biggest improvement.
Stability greatly improved. All AI-engine related issues should be resolved.
Improved Windows GPU compatibility for weaker graphics cards, and Mac compatibility for Mac devices released before 2012.
New pop-up after installation that includes key information for processing. This same information can be accessed by going to the Help menu. Fixes Minor GUI refinements and bug fixes
1.2.1 (17/04/2020)
Major features
Artemis models updated. Artemis-MQ has been removed due to having consistently similar results to Artemis-LQ.
Artemis OpenVino support enabled for CPU use only.
Improvements
Installers will now fully clean out the installation directory before proceeding with the installation.
Output file permissions will be verified during the rendering process.
GPU command line functionality implemented
Fixes
Increased general stability and applied numerous bug fixes
1.2.0 (25/03/2020)
1.2.0rc идентична 1.2.2beta
Models update:
To reduce future confusion, we use Greek goddess names for model types: the previous "HQ" & "HQ-CG" models are named "Gaia", as "Gaia-HQ" and "Gaia-CG". The previous "LQ" model is named "Artemis" and split into 3, "Artemis-HQ/MQ/LQ" to use with input video of high/moderate/low quality. More info can be found in the tooltips.
GUI refinement:
Display second per frame during processing/preview.
Add the ability to directly enter the current frame, the start frame, and the end frame indices.
When you click on the right-top red "X", there will be a warning message.
you can use menu: process|preference to set AI VRAM(GPU)/RAM(CPU) usage. The high number tends to make it faster. Lower the number if it is unstable.
1.1.1
Add option to keep audio in output .mp4.
Bugfixes
В окне "About" по-прежнему указана версия 1.1.0
1.1.0
Introduced 100% presets for all the 3 models. These presets can handle worse noise and block artifacts than other presets. Please note that they are designed for digital camera noise reduction, so grain generated by scanning may not fit for these models. You can also trigger them in custom settings by setting the scale < 120%.
Updated all the 3 models for quality improvement. (добавлены модели 1х)
Fixed a file reading bug.
Now after preview loop stops, the frame cursor will go back to the start point
Model name (HQ, HQ-CG, LQ) now is included in the output name
Minor bugs fixes
1.0.2
Just release
fixed nearly all the major issues
many minor refinements
RC1.0.1 (08/02/20)
Fixed a bug that slows down processing
Improved login/purchasing process
Many minor refinements
Disable output audio (we are in the process of switching a better library. We have trouble to sync audio/video in the current version)
RC1.0.0 (07/02/20)
Refinements & updates:
Updated all the 3 models for quality improvement
Audio is merged back to output video
New icons
Trimmed unnecessary files and now the whole folder will occupy smaller space on your disks Bugs fixed:
Fixed drag&drop failure (after batch processing) bug
Тихая\скрытая установка:
Код:
foo.exe /S
foo.exe /S /D=C:\Program Files\Foo
Ключи чувствительны к регистру. Путь установки должен быть в конце и без кавычек.
Beta-версии
1.5.0.2 beta (21/08/2020)
We have a new beta for you to test. This beta is testing support for some of the backend changes needed to be made to support the features in the alphas. We are going to be rolling out those new features in the coming days into betas, but for now we need to test to make sure none of the changes made so far break anyone's setup so *please* take some time to install and at least test 1 video with this version especially for windows users.
We did fix some bugs in this version:
Choose between High bit rate H.264 or Low bit rate H.263 (yes we realize adjustable H.264 bit rate would be ideal and we are working toward it, but this should be a good stop gap solution for those needing to use less disk space.
Fix for saving default model/params
Loading unlocked aspect ratio videos now works
Update to the latest compiler version to support the new a.i. engine on windows (New engine to be released soon. Currently in Alpha)
Fix for model tooltips showing out of order.
Please upload your crash reports here: https://forms.gle/9zAKDt2UF3KJNKov9
We read and respond to every crash report.
WIN: https://downloads.topazlabs.com/packages/win/videoenhanceai/VideoEnhanceAI-1.5.2....ws-installer.exe
MAC: https://downloads.topazlabs.com/packages/mac/videoenhanceai/VideoEnhanceAI-1.5.2....sx-installer.dmg
1.5.0 beta (03/08/2020) TensorCore & AMD GPUs support
Windows: https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.5...ws-installer.exe
MacOS: https://topazlabs.s3.amazonaws.com/packages/mac/videoenhanceai/VideoEnhanceAI-1.5...sx-installer.dmg
On the surface, this version is exactly the same as the previous one. There are no new feature additions.
Under the hood, we changed some things internally so that we can support Tensorcore and also AMD GPUs. (in the future, AMPD support is still not turned on)
We would like to test this change to make sure all Windows machines can work with this configuration. If you experience any instability, please remember that this is a test of a new framework, and we are very interested in the results.
Please upload your crash reports here: https://forms.gle/9zAKDt2UF3KJNKov9
We read and respond to every crash report.
On the MacOS we have a made a change so that your machine will use all available GPUs, including eGPUs that you have plugged in. The ability to turn on and turn off specific GPUs is coming in a future release. Changelog:
Windows
Internal changes to support TensorCore and AMD GPUs (Not activated).
Mac:
Multiple GPU support, now VEAI will use all available GPUs including eGPU. GPU selection will be added in future version.
1.4.1 beta (23/07/2020)
We have a new Beta 1.4.1
Mac: https://topazlabs.s3.amazonaws.com/packages/mac/videoenhanceai/VideoEnhanceAI-1.4...sx-installer.dmg
Windows: https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.4...ws-installer.exe
To address the elephant in the room we have added back in mp4 output. We have also decided to temporarily remove the post-processing from the GUI. We just weren't ready to release it to the public yet. If you are on Windows though you can re-enable it using the steps at the bottom of this post. However we won't be providing any support for it for now.
I also wanted to give a bit of a roadmap update as far as output goes. We are currently working on overhauling our entire video (and image) I/O. This will bring color space improvements, stability improvements, and move us from H.263 to a more modern format 😊 for direct video output. We also hope to address the audio sync issues. This does take time though so we appreciate your patience. That said we are not currently planning to turn VEAI into Adobe Media Encoder with a plethora of output options. We hope to be able to support most people with a combination of a H.264/5 option for people who want to get a video they can use directly and then some "intermediate" formats for our users who want to re-import the videos into other programs for further processing (EXR, TIFF, PNG, JPG, ProRes, etc. are all being considered). The post processing was a bit of a stop-gap solution and not fully baked. We may at some point return to it and use a similar methodology but not process all the frames at once and instead process them in chunks so that we can conserve disk space and improve usability.
Thank you all for taking the time to test our software and we look forward to hearing what you think of the latest build. Changelog:
Image Sequence Input for the Command Line
Network paths on Windows for the Command Line
Theia Model update
Removed post-encoding and restored output options to previous version. Instructions to re-enable post-processing:
On Windows:
Open the Registry Editor
Go to the path "Computer\HKEY_CURRENT_USER\Software\Topaz Labs LLC\Video Enhance AI (BETA)"
Create or set a String Value called "postProcessingEnabled" and set it to "true"
VEAI 1.3.6 RC(release candidate) is ready here:
WIN: https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.3...ws-installer.exe
MAC: https://topazlabs.s3.amazonaws.com/packages/mac/videoenhanceai/VideoEnhanceAI-1.3...sx-installer.dmg Here is the changelog:
- Updated Theia model
- Now for machines with Mac OS 10.15 or above, Artemis models will run on GPU
- Stability improvements for windows
- Custom output directory
- Custom image names for output
- More detailed tool tips and messages
- Fixed issue with mac version number Known issue:
- Keeping audio will sometimes cause problems when output is set as mp4 on some MACs.
This version is the release candidate of the coming release. We will be more then thankful if you can help us to test it intensively, as you always did. Please enjoy it and the coming holiday!
1.3.5 beta
VEAI 1.3.5 beta is ready here:
WIN: https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.3...ws-installer.exe
MAC: https://topazlabs.s3.amazonaws.com/packages/mac/videoenhanceai/VideoEnhanceAI-1.3...sx-installer.dmg
Here is the changelog:
- Speed improvements up to 30% (Win and Mac) for large size videos.
- Fixed the cropping bug.
- Theia models: updated with limited anti-aliasing ability (for line-skipping cameras).
- Theia models: improved image quality and motion consistency.
- Up to 7 digits for frame number.
- ETA on frames per second is more accurate.
- Preview can be panned by dragging. Mouse wheel and Cltr-+/- key change zoom.
- Preview can step forward and backward using left/right arrow keys.
- Allow importing same video multiple times so that it can be processed with different options.
- Default output video name will add "(1/2...)" to the end if it exists to prevent from being overwritten.
- In video list, replace the check mark with a checkbox so that user can skip videos (uncheck them) at batching.
- Delete all video pops up a warning dialog.
https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.2...64-installer.exe
Compared with 1.2.5 beta, this version should have nearly all the AI Engine failure issues fixed. And we temporarily rolled back the h264 codec
to mpeg4 codec (so now we support mpeg4 & vp9). More codecs and refined writer are on the roadmap.
Please note that for WIN users, if you have updated your Nvidia driver, you need to restart your computer to make it work. And because that we update the libs, please do not run this beta together with the release version. So our suggestions are:
1). Update your NV driver if you haven't do so;
2). Restart your computer;
3). Run this beta and do NOT run any release versions.
Another thing we forgot to mention in the 1.2.5 beta post is that we cannot support any NV cards with <4GB VRAM. We are sorry but this is what we need for the minimum speed and quality requirement.
1.2.5 beta (25/04/2020)
https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.2...64-installer.exe New features:
Updated Windows and MAC libs. Now WIN Nvidia GTX 9xx series cards and older MAC hardware up to 2012 are supported!
VP9&h264 video output codecs are supported. There is an option on the right panel for it. Bug fixes and refinement:
ETA is more explicit in the format of days:hours:mins:secs.
Fixed a missing frame bug.
Cleaned up unused libs in MAC version.
Please note that for WIN users, you may need to update your Nvidia driver for this version since the opencv, cuda and tensorflow libs are updated. If you have any touble with this version, please let us know and we will look into your issues.
Removed Artemis-MQ model since it would generate nearly the same output as Artemis-LQ model.
Speed improvement:
We integrated Intel toolkit for the Artemis models. No you can see about ~2x speed up in the CPU mode.
Bug fixes and refinement:
Now if your output video file is read-only, VEAI will stop saving and pop up a message.
GPU selection available in cmd line options. Please check with "--help" for more details.
We fixed some bugs that will cause VEAI crash.
1.2.2 beta (24/03/2020)
https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-1.2...64-installer.exe
Major Changes:
fix some stability issues. Please let us know if it still crashes on you.
you can use menu: process|preference to set AI VRAM(GPU)/RAM(CPU) usage. The high number tends to make it faster. Lower the number if it is unstable.
Updated the Gaia-HQ model to reduce texture pattern.
Updated all Artemis models with more training.
Please let us know if there are any issues. If no major problem, it will be the last beta before an update release. Again stay safe and happy upscaling.
To reduce future confusion, we use Greek goddess names for model types: the previous "HQ" & "HQ-CG" models are named "Gaia", as "Gaia-HQ" and "Gaia-CG". The previous "LQ" model is named "Artemis" and split into 3, "Artemis-HQ/MQ/LQ" to use with input video of high/moderate/low quality. More info can be found in the tooltips.
GUI refinement:
Display second per frame during processing/preview.
Add the ability to directly enter the current frame, the start frame, and the end frame indices.
When you click on the right-top red "X", there will be a warning message.
We also separate the beta version from the release version. Now the beta will be installed in a different folder so you don't have to worry that it will mess up the release version.
Even the program looks the same, there are some major code changes to position us for the next stage of development, especially more models, better speed improvement, and more GPU support. On the Windows side, CPU and Gaia models may run slower than the release version. This is temporary. GPU speed is not affected.
Не поддерживается кириллица в имени файла и длинные имена.
Другие продукты и версии программ от Topaz Labs Предыдущая консольная версия Topaz Gigapixel AI for Video 0.1.8Обновлено до RC1.0.1 (08/02/20) Обновлено до 1.0.2 (12/02/20) Релиз Обновлено до 1.1.0 (04/03/20) Обновлено до 1.1.1 (10/03/20) добавлен звук Обновлено до 1.2.0rc (25/03/20) Обновлено до 1.2.1 (22/04/20) Обновлено до 1.2.2 (07/05/20) Обновлено до 1.2.3 (07/06/20) Обновлено до 1.3.8 (09/07/20) добавлены модели Theia Обновлено до 1.4.2 (30/07/20) Обновлено до 1.5.1 (13/08/20) h.264 Обновлено до 1.5.2 (27/08/20) h.264
По большому счету, прога делает раскадровку и увеличивает кадры с теми же алгоритмами, что и Gigapixel Ai. В финальном видео видны дикие артефакты.
На Gtx1070 работает.
78837868По большому счету, прога делает раскадровку и увеличивает кадры с теми же алгоритмами, что и Gigapixel Ai. В финальном видео видны дикие артефакты.
Во-первых, это не алгоритмы а модели. Во-вторых, модель в этой версии заметно отличается от обычного GP. Она намного лучше понимает артефакты, связанные именно с видео. К тому же, HD-CG-модель даже близко не имеется в обычном GP. Необходимость в ней правда не столь велика, разве что кто-то хочет увеличить геймплейные ролики для заливки на ютуб в повышенном разрешении и избавиться от рябящих лесенок.
Программа установила. при запуске ничего не потребовала.
Win 10.
при открытии файла 720х576 dv avi pal 1 минута, программа вылетает.
при открытии кодека h264 720х576 длительностью 1 минута также вылетает.
так и должно быть?
проц i7700k
видео ASUS GeForce GTX 660 .
78852714Программа установила. при запуске ничего не потребовала.
Win 10.
при открытии файла 720х576 dv avi pal 1 минута, программа вылетает.
при открытии кодека h264 720х576 длительностью 1 минута также вылетает.
так и должно быть?
проц i7700k
видео ASUS GeForce GTX 660 .
Смотрите внимательно шапку
Цитата:
Минимальные системные требования:
• Только Windows 10 x64
• Только Nvidia
• GTX 1050 with 4 GB VRAM !!!
• либо iGPU (процессор со встроенной графикой)
Рекомендуемые требования:
• 32 GB RAM
• nVidia graphics card with 6GB or more VRAM
с GTX 660 работать не будет точно. Как вариант, если отключить видеоплату(лучше аппаратно), то возможно прога заработает на встроенном графическом ядре процессора i7700k. Но даже если прога запуститься все будет крайне медленно.
Появилась финальная версия, которую можно скачать по ссылке (триал) или купить с половиной стоимости. Вся информация под спойлером. На официальном сайте страницы продукта пока нет, но купить можно по ссылке ниже.
скрытый текст
We finally have our public release version of Video Enhance AI here: https://topazlabs.s3.amazonaws.com/packages/win/videoenhanceai/VideoEnhanceAI-win...64-installer.exe
We fixed nearly all the major issues you reported with many minor refinements.
Thanks for all your help and feedback! As a token of our appreciation, here is a special coupon code for you:
veai99
You can use the code to buy Video Enhance AI at $99.99 (list price $299.99). Please note that this code will expire by the end of this month. We trust you not posting it anywhere outside this group. Here is the direct link to purchase: https://topazlabs.com/cart/?add-to-cart=1464951
Thanks again and please enjoy it!
P.S. We have some trouble mass resetting trial dates. If you still need another 30 days’ trial, please send an email with your Topaz Labs account email to cloudai@topazlabs.com.
А также появилась возможность бесплатно получить все продукты Топаза, кроме Video Enhance AI (он же Gigapixel AI for Video). Порядок действий также под спойлером
до конца февраля 2020
Заходим на официальный сайт топаза
Переходим в раздел Products -> AI Bundle (для добавления сразу всех программ на аккаунт), либо на страницу конкретного продукта, если вам нужен только он.
Добавляем в корзину сначала The Utility Bundle, затем возвращаемся и добавляем The Creator Bundle.
В корзине вводим купон veai99 и получаем 100% скидку (на всё, кроме Video Enhanced AI).
Нажимаем Proceed to checkout
Вводим почту существующего аккаунта, либо (если аккаунта нет) любую другую рабочую почту, на неё придёт пароль от аккаунта.
Далее вводим любые данные (они не проверяются автоматически) Теперь в самой программе вводим почту, которую вы указывали, и пароль от неё (если аккаунт новый, то пароль на почте).
Профит.
Обновлено до релизной 1.0.2
вылечено. Если попросит логин\пароль - просто жмём ОК. ~Razdor~
Я чо спросил-то про developer.nvidia.com/rtx/ngx
Насколько я понимаю у них скачать можно только SDK для разработчиков. А готового решения с использованием NGX не встречал?
magkms
Я неделю назад с NVIDIA NGX SDK попытался сработаться. Последняя версия 1.1. У них есть готовые примеры для компиляции. Скопилировал DLVSR (тот что для теста технологии AI апскейла ) на Visual Studio 2019 . Что хочу сказать, скорость колосальная, на nvidia RTX 2060 с 1080p до 4К процессинг в среднем порядка 8,5 кадров в секунду. Вот только результат по моему мнению еле заметный (если сравнивать с Topaz AI Gigapixel то второй в разы выигрывает). Возможно улучшат в будущем, AI модель там очень слабая, нужно обучать. Но возможно обучать ее пока разработчики не предусмотрели, но планируется в ближайшем будущем.
Если кратко, то пока все еще сыро, и эффект минимален. Но возлагаю большие надежды именно на NGX.
По последней версии - меняет немного цвет видео, в зелёный оттенок. Легко прослеживается прямо в проге, а на отрендером видео тем более. Звук так и не преносит, что печалит.
magkms
Огромное спасибо за лайфхак! Скачал, зарегил. Интересно, как долго можно будет пользоваться? Неужто не обнаружат, что куча людей ломанулась халяву ловить на один купон?
78879624magkms
Огромное спасибо за лайфхак! Скачал, зарегил. Интересно, как долго можно будет пользоваться? Неужто не обнаружат, что куча людей ломанулась халяву ловить на один купон?
Это же купон а не аккаунт с логином/паролем от чужого аккаунта...
78877430Никто не знает сколько жрет видеопамяти DAIN для рендера FullHD? Опытным путем установлено что 8гб тянет только 480р что очень печально(
Да, увы около 500 пикселей это пока что максимум для этого проекта. Но он вроде активно развивается и там решают эти проблемы, так что думаю в течении года проект выйдет на нормальный уровень и можно будет рендерить ФуллХД
На GTX 750 Ti, конечно же, не работает.
Процессор грузится до отказа, на превью выдаёт оригинальное видео со скоростью 1 кадр за полторы минуты. Гигапиксель при этом работает отлично.
