FutuTek

Пікірлер

@DavidTran-j1z Күн бұрын

Thanks very much for your workflow. @FutuTek. The size of Reese_Finch.safetensors is right only 19KB? And the result is very weird with your result. I think some models are differ from you. Plz lemme know the correct model name and links. Thank again.

@murtazaeng 3 күн бұрын

I have a GTX graphics card, a large workflow like this will not work on my PC , I would expect a simplified version, without the enhancements, I was always searching for ageing videos tutorials.

@aegisgfx 7 күн бұрын

guys really need to get out there and meet real wimmin lolz

@markdkberry 8 күн бұрын

ran into problem after problem trying to run this unfortunately. I think some updates may have broken it. now getting "RuntimeError: torch.cat(): expected a non-empty list of Tensors" and it hangs on batch prompt schedule #78 but its setup as per your download and the video. just before that error was "WARNING: Sequential prompt keyframes 20:307 and 21:247 are not monotonously increasing; skipping interpolation."

@FutuTek 7 күн бұрын

Hi, I recommend you to full update your comfyui, then try to run just 1second (set 24 in Frame load cap at node #143) for testing

@markdkberry 8 күн бұрын

wow. this is some genius workflow. I am just trying to run it now. I want to adapt it to help me fix hunyuan poor qual characters as my 3060 12GB Vram potato can only handle small size vids using the hunyuan GGUF.

@OfficialSidelineSecrets 9 күн бұрын

great video, just lose the music or lower the volume its annoying

@sexybbunny 9 күн бұрын

klingAi is a page, can we get results like that with comfyui?

@sexybbunny 9 күн бұрын

I'm testing your flows and watching your videos, I want to get the same results, please keep posting, I'm inspired by your videos

@abhranilghosh77 11 күн бұрын

which gpu you are using, and how much time does it take to process

@FutuTek 11 күн бұрын

I use a RTX 4090 and it took around 45 min

@abhranilghosh77 10 күн бұрын

@@FutuTek i have RTX 4060 laptop gpu and at advanced ks sampler torch is out of memory.

@FutuTek 10 күн бұрын

@@abhranilghosh77 You can reduce the Upscale resoution in group node 1 lower, example 1024x576, and shorten the video to under 10s, frankly you should not use laptop for AI running otherwise you can burn it ^^

@MrCalmwaters112 21 күн бұрын

Can't hear anything. Why would you have the music so loud?

@quangdinhdota2388 22 күн бұрын

Thank for your video. Can you note detail file for each box in step 2. Same file I don’t know how I should add exactly file fill in box. Name file different in your video ❤

@QuanticalCapybara 23 күн бұрын

Nice workflow ! I'm playing with it a lot and it's really great ! Thanks for sharing it

@FutuTek 22 күн бұрын

Nice to hear that

@QuanticalCapybara 23 күн бұрын

You're a AI hero ! Your work is fantastic. Thanks for sharing it 😘

@mohammedzaidkhan5687 27 күн бұрын

How much time did it take to convert the video you showed us, Please respond, i would appreciate it brother

@FutuTek 26 күн бұрын

I use a RTX 4090 GPU then it took me 30 min

@mohammedzaidkhan5687 26 күн бұрын

@FutuTek thanks a lot brother

@jorgediaz9093 27 күн бұрын

Lower the background music is to loud to annoying

@ian2593 29 күн бұрын

Great workflow! This is the first one that has applied the style without artifacts for me. I'm wondering how many frames it can convert on a mid range machine with an RTX 4070 series.

@FutuTek 29 күн бұрын

You should convert the max 15s video with 12fps, then use Film FVI to improve the interpolation to upgrate to 24fps

@waynegilbert9517 29 күн бұрын

I got an error: OpenPoseXL2.safetensors' not in (list of length 24) in the "Hi-Res Fix Script" Can you tell me how to fix it?

@waynegilbert9517 29 күн бұрын

Got it running after some research

@APP-STOLZ 4 күн бұрын

Make sure the file 'OpenPoseXL2.safetensors' is in the correct model folder and that the file path is properly set in your Hi-Res Fix Script. If it's missing, download the file again from the source.

@larryleng8531 Ай бұрын

我想如果利用這個工作流也可以製作從種子到開花到果實到凋謝的全過程

@FutuTek Ай бұрын

Of course, it it is the same method and logic

@larryleng8531 Ай бұрын

我們一直嘗試在讓AI既能發揮他的能力又需要規範他的動作，所以我們不得不把更多的控制也就是約束加給他，這讓我很迷惑到底是我們在創造AI還是AI在探查我們的底綫。這太難了另外這個工作流的消耗確實是驚人的，我的頂級設備，在運行的時候遇到了兩次溫度報警。但這依然是給我非常多的啓發，非常感謝你的貢獻我愛你，最誠摯的尊重，來自東方的祝福。

@FutuTek Ай бұрын

You should reduce the video resolution and shorten the video to avoid the high temperature

@larryleng8531 Ай бұрын

偉大的創舉，我一直在嘗試畫面的協調的問題，畫面和動作以及風格，你做到了，我一直想簡化流程單事實證明這沒有捷徑必須一點一點的處理細節，你的控制部分太棒了給了我很多的啓發，應該還能更高，非常感謝你。

@MisterWealth Ай бұрын

My footage just comes out discolored/purple for some reason, do you know why?

@FutuTek Ай бұрын

Do you mean you use my footage in description? or you own footage? if you use your own footage please try to re-create it with higher frame rate

@MisterWealth Ай бұрын

@@FutuTek My own footage, for some reason if its not slow motion video, itll have a weird purple cast, or it won't detect the faces, or generally have some terrible artifacts

@KanaNoClown Ай бұрын

So good

@nholmes86 Ай бұрын

Superb conversion

@ivanbelyakov Ай бұрын

Thanks for the video! I need help. It issues errors, and then the calculation stops at 9%... lora key not loaded: lora_unet_up_blocks_0_attentions_1_transformer_blocks_0_attn1_to_k.lora_down.weight ERROR lora diffusion_model.output_blocks.3.0.in_layers.2.weight shape '[1280, 2560, 3, 3]' is invalid for input of size 11059200

@ivanbelyakov Ай бұрын

changed LCM SDXL to LCM 1.5 and the process started. Then it hung at 9%, with 95% GPU load. I waited 5 hours, but nothing happened. The scheme and example files are yours.

@FutuTek Ай бұрын

@@ivanbelyakov There is not enough RAM (not VRAM), you need a computer at least 32G RAM, then please reduce the upscale resolution to 1024x576 then run test for 1 second (24 frame only), then increase it gradualy

@ivanbelyakov Ай бұрын

@@FutuTek I have 64gb RAM

@FutuTek Ай бұрын

@@ivanbelyakov I have 64G too, and sometiem the same problem still occured, just reduce the resolution or shorten the length then the problem will be solved

@quangdinhdota2388 Ай бұрын

Same problem. Have you solved problem?

@Djonsing Ай бұрын

NVIDIA GeForce RTX 2060 SUPER (8 GB) does not pull, writes not enough memory

@FutuTek Ай бұрын

You need a GPU at least 12GB bro

@QuanticalCapybara 23 күн бұрын

@@FutuTek I have a 3070ti 8gi and it works well. I lowered the video output for testing purpose

@mrrubel8841 13 күн бұрын

@@FutuTek 3060 will work?

@miratdrati Ай бұрын

Can I change the video style bro? where to change, I'm new to comfyui

@FutuTek Ай бұрын

What you mean "video style"?

@miratdrati Ай бұрын

@ yes. i want change style video, eg : chibi ?

@FutuTek Ай бұрын

@@miratdrati you just go to civitai.com, type chibi to the search, enter, then select model SD 1.5 on the left, there are many chibi model for download, just download to comfyui\models\checkpoints, then refresh the workfow. After that you can choose new model in the Group Node 2 (load checkpoint)

@Pepe-kf2gj Ай бұрын

hola, gracias por tu trabajo, cuanto de vram se necesita para que funcione

@FutuTek Ай бұрын

Resolution 576x1024: at least 12G Resolution 720x1280: 24G

@FutuTek Ай бұрын

Because many ones asked, I inputed all models and loras links to the description, let's have fun

@dmitrykonovalov9366 Ай бұрын

I searched and found most but not all loras/vae/models. It would be nice if you could add the links to the safetensors you are using.

@FutuTek Ай бұрын

I inputed all models and loras links to the description, let's have fun

@dmitrykonovalov9366 Ай бұрын

Nicely explained! Many thanks.

@mayazus1 Ай бұрын

Question: I get a red box around HighRes-Fix Script. I don't know what wrong with it. Settings are set to: Laten, (use same), ttl_nn.sd 1.x, True, 8, .75, 1, false

@FutuTek Ай бұрын

First you should update all the comfyUI Then, did you connect it to "script" of Ksamplet? if not you can try by right click to the node and select Fix node (recreate) The last solution is that you deleted it and double click to any place, type "Highes-Fix" to the Seach Node and newly create it

@markdkberry 8 күн бұрын

@@FutuTek I have same issue. everything else works. tried all you suggested. it said a bunch of stuff "not in list" first time but deleted and recreated it and now it goes red and shows this in terminal: "Failed to validate prompt for output 128: * HighRes-Fix Script 274: - Value -1 smaller than min of 0: seed Output will be ignored" will keep with it see if I can figure something out. havent got past it yet but I think my animatediff was wrong last go.

@markdkberry 8 күн бұрын

@@FutuTek I fixed the "highres-fix script" issue, you have to set "use same seed" to false, then change the number from -1 to 0. but I now have it crashing with efficient loader: "header too large" EDIT: the error was in #185 node not the ksampler. I have the wrong chkpt model in that. EDIT 2: no luck, still getting #185 saying "effcient loader" header too large. EDIT 3: all good, one of the connecting loras was 1kb in size must have downloaded it wrong.

@sprocketamerica1 Ай бұрын

Getting: Prompt output failed validation: Exception when validating node: name 'full type name' is not defined VHS_VideoCombine

@FutuTek Ай бұрын

You can try by right click to the note and select Fix node(recreate)

@sprocketamerica1 Ай бұрын

@@FutuTek no sorry does not fix it.....please review your work flow you posted because i think it is differnt then you video

@FutuTek Ай бұрын

@@sprocketamerica1 I have just added a new workflow to the folder, name "...241213", I already tested it, if the problem still occur pls. check your software and computer setting

@sprocketamerica1 Ай бұрын

@@FutuTek what version of comfyUI are you running this may be my issue because i still get he (Prompt outs failed validation) Exception when validating node : name 'full type name' is not defined VHS_VideoCombine

@adukomik720 Ай бұрын

Hi Fututek! I managed to install all nodes and followed your guide and I have quite some appealing results. How do I fix faces since they keep changing and male characters look so feminine. find link to what I achieved following your guide, thanks. drive.google.com/file/d/1XeKkRRYSQvcxUnwipIuSZhXStALo3vKh/view?usp=sharing

@Isi-uT Ай бұрын

Nice!

@adukomik720 Ай бұрын

Super awesome! and thanks for the workflow. However I am missing some nodes and being a newbie on comfyui I don't know the workaround. When you find time please help. Thanks

@FutuTek Ай бұрын

Hello please follow below steps: 1. Update your ComfyUI to the latest version: Manager / Update All, then restart ComfyUI 2. Load my workflow in the video discription 3. Installing missing notes: Manager / Install Missing Custome Nodes, then have fun

@adukomik720 Ай бұрын

@@FutuTek I appreciate you for responding. Yes, I have updated to the recent comfyui, still figuring out how to fix import failed ComfyUI-Advanced-ControlNet and AnimateDiff Evolved. Thanks

@adukomik720 Ай бұрын

@@FutuTek what models did you use. SDXL or v1.5?

@FutuTek Ай бұрын

@@adukomik720 Try to install them separately in Manager / Custom Nodes Manger

@FutuTek Ай бұрын

I use SD 1.5, the SDXL is not much good for AnimateDiff

@FutuTek Ай бұрын

Hello, it is very easy, you load my workflow to your ComfyUI, then you will see the "modelname.safetensor" or "modelname.ckpt" in every workflow, just type this full "modelname.safetensor" to Google and then you will seed a lot of sources to download it

@ComfyUIstudio Ай бұрын

Hi, love your video style. can you add to the workflows model links in a note node?

@FutuTek Ай бұрын

@CalladsEssence Ай бұрын

Really looking great! Also thank you for the workflow! 💋

@FutuTek Ай бұрын

Thank you & glad to hear that

@sextoyultimo6177 Ай бұрын

What GPU do you use? The problem I always have with every animation setup I've tried from ComfyUi is consumption. I have an RTX 3060ti and only in ComfyUi do I have consumption problems. I don't know if I have to have a very specific workflow or if it's already intrinsic to the SD version that it consumes more by default regardless of which workflow is used. I say this because in the WebUi version some things do not consume me as much as in ComfyUi, but as I say, I don't know if this is more a problem of not knowing how to make an adequate workflow or if it is something that comes by default in the version. and a more powerful GPU is necessarily required to work in ComfyUi, no matter what.

@FutuTek Ай бұрын

I am using a 4090 bro, you can set lower resolution if you use a lower GPU

@jackh1749 Жыл бұрын

Its super mattress man!!! 😂😂😂

@ronnbaker7775 Жыл бұрын

I'd love it if the music actually keep forgetting the barkboard thing

@kennethwilliams7246 Жыл бұрын

The Volocopter is just an electric helicopter, underwhelmed.

@stevewiggin1062 Жыл бұрын

The future is here among us... .,.

@Ekamisgoated Жыл бұрын

Bros opps is iron man☠️

@josephperreault2063 2 жыл бұрын

The cold magnet 🧲 is perfect is it had back fans to thrust it forward

@josephperreault2063 2 жыл бұрын

Well it’s a start with all the wrong parts because I have theory I’m such a genius

@Kwago1 2 жыл бұрын

If any of the planets in the solar system were to have less of a gravitational pull, it wouldn't be someplace bigger than the one we're on. More efficient for the smart lab tech thought the price for more use of material over another overused planet outcome. Now drowning as we speak.

@epicvillain8308 2 жыл бұрын

Lol there are no limits, except going higher than 5 inches 😂

@connorrichardson5691 2 жыл бұрын

Igen, igazad van az oldaladon található cikk informatív volt, és megakadt a szemem, remélem nem bánjátok, ha jobban megismeritek egymást.

@jslevenson101 2 жыл бұрын

Gyroscopes might help.

Ең жақсы KZbin

Пікірлер