video2x
video2x copied to clipboard
k4yt3x
GUI fixed -hwaccel to auto
Environment Information
Version Video2X 4.3.0
FFmpeg not working with Nvidia GPU using -hwaccel auto and in GUI or YAML file if you change parameter to -vsync 0 -hwaccel cuvid not working and continue forcing to -vsync 0 -hwaccel auto, which use CPU to extraction and assemble, if possible, put a field to force -vsync 0 -hwaccel cuvid for example to force use nvidia GPU in FFmpeg.
You can change the method here in the config file.
BTW, could you please break sentences into smaller sentences the next time for readability? Also the typos.
You can change the method here in the config file.
BTW, could you please break sentences into smaller sentences the next time for readability? Also the typos.
Really sorry I am not be able to be more precise, my english is very basic, but I prefer this, otherwise use translator and change too much the sense of the phrase. I had try do the change in config file before open the issue and not sucefull with this, for any reason, the code is change to auto, how can you see in this images.
So, no mather what the program in auto never use nvidia gpu in ffmpeg just the processor, if you fix this, the conversion will be more faster using a nvidia gpu.
I am not a programer in python, so I not dare try chante your code, but I think the problem is in video2x_gui.py
in line
I think I discovery another important thing, because I am trying to do a runing in colab google for linux. If you use the parameter -pix_fmt rgba64be wich is default, not be able to run with -hwaccel cuvid, this parameter -pix_fmt rgba64be is needed or not??
If you are interested in see my cloud runing of the software, follow the link, but is a work in progress. https://colab.research.google.com/drive/1E_CexJNJfjDV36fWySDOek38k8MdoVsI?usp=sharing whit this the users can try the software whithout nvidia gpus or big cpu processing.
- What GPU are you using?
- ffprobe output of the video (located at the Tools Tab)
- What GPU are you using?
- ffprobe output of the video (located at the Tools Tab)
The gpu is a GTX 1060
The second information I will try get next run, but I think you don´t understand, the option are overwrithed by the software, does not mether if I set cuvid, It changes to auto. in config file or not.
set cuvid, It changes to auto
That could be a bug but that's most likely not what's affecting you in the first place (auto should pick cuvid if available).
Your video codec might not be compatible with hardware acceleration, hence its falling back to software decoding.
set cuvid, It changes to auto
That could be a bug but that's most likely not what's affecting you in the first place (auto should pick cuvid if available).
Your video codec might not be compatible with hardware acceleration, hence its falling back to software decoding.
That is my point, is a bug, because does not mather if I set cuvid the gui program change to auto and not use the gpu to decode ffmpeg. I use last version of vide2x and update my nvidia drivers and issue is the same.
My Linux FFmpeg will use 'cuda'(not cuvid) when using '-hwaccel auto'. My GPU is GTX750Ti
I am not a programer in python, so I not dare try chante your code, but I think the problem is in video2x_gui.py in line
I see what's going on there. I might consider adding a drop down menu or a text box so you can choose the hwaccel value.
- What GPU are you using?
- ffprobe output of the video (located at the Tools Tab)
Follow the fprobe
"streams": [ { "index": 0, "codec_name": "h264", "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10", "profile": "Constrained Baseline", "codec_type": "video", "codec_time_base": "1001/60000", "codec_tag_string": "avc1", "codec_tag": "0x31637661", "width": 400, "height": 300, "coded_width": 400, "coded_height": 304, "closed_captions": 0, "has_b_frames": 0, "sample_aspect_ratio": "1:1", "display_aspect_ratio": "4:3", "pix_fmt": "yuv420p", "level": 21, "chroma_location": "left", "refs": 1, "is_avc": "true", "nal_length_size": "4", "r_frame_rate": "30000/1001", "avg_frame_rate": "30000/1001", "time_base": "1/30000", "start_pts": 0, "start_time": "0.000000", "duration_ts": 9290281, "duration": "309.676033", "bit_rate": "581194", "bits_per_raw_sample": "8", "nb_frames": "9281", "disposition": { "default": 1, "dub": 0, "original": 0, "comment": 0, "lyrics": 0, "karaoke": 0, "forced": 0, "hearing_impaired": 0, "visual_impaired": 0, "clean_effects": 0, "attached_pic": 0, "timed_thumbnails": 0 }, "tags": { "creation_time": "2020-04-13T23:15:55.000000Z", "language": "und", "handler_name": "ISO Media file produced by Google Inc. Created on: 04/13/2020." } }, { "index": 1, "codec_name": "aac", "codec_long_name": "AAC (Advanced Audio Coding)", "profile": "LC", "codec_type": "audio", "codec_time_base": "1/44100", "codec_tag_string": "mp4a", "codec_tag": "0x6134706d", "sample_fmt": "fltp", "sample_rate": "44100", "channels": 2, "channel_layout": "stereo", "bits_per_sample": 0, "r_frame_rate": "0/0", "avg_frame_rate": "0/0", "time_base": "1/44100", "start_pts": 0, "start_time": "0.000000", "duration_ts": 13660160, "duration": "309.754195", "bit_rate": "96001", "nb_frames": "13340", "disposition": { "default": 1, "dub": 0, "original": 0, "comment": 0, "lyrics": 0, "karaoke": 0, "forced": 0, "hearing_impaired": 0, "visual_impaired": 0, "clean_effects": 0, "attached_pic": 0, "timed_thumbnails": 0 }, "tags": { "creation_time": "2020-04-13T23:15:55.000000Z", "language": "und", "handler_name": "ISO Media file produced by Google Inc. Created on: 04/13/2020." } } ], "format": { "filename": "E:\video\raul.mp4", "nb_streams": 2, "nb_programs": 0, "format_name": "mov,mp4,m4a,3gp,3g2,mj2", "format_long_name": "QuickTime / MOV", "start_time": "0.000000", "duration": "309.754000", "size": "26313070", "bit_rate": "679586", "probe_score": 100, "tags": { "major_brand": "mp42", "minor_version": "0", "compatible_brands": "isommp42", "creation_time": "2020-04-13T23:15:55.000000Z" } } }
My Linux FFmpeg will use 'cuda'(not cuvid) when using '-hwaccel auto'. My GPU is GTX750Ti
I try this 'cuda' and the issue is the same, by the way I am using windows.
Fixed in #373 Although due to the settings being used with ffmpeg there is a negligible speedup as all of the options use the CPU to do the encoding rather than the GPU see: #374
