So this is kind of out of desperation but maybe someone will save me. Do you have a solution for scaling the simulation? At scale 1 yeah the simulation runs perfectly but at .001 the size, the simulation screws up. Is there any way to have the simulation calculate everything as if it was at the scale of 1 but be at the scale of .001? Maybe some kind of instancing I am missing perhaps?
Thank you again. I need to link it with unreal but I did not know any video tutorials for this as you do it .. Please step by step to follow you with my best thanks
I am using metahuman and every time I set up my collision in the physical assets, applied it to my metahuman blueprint, it worked, until I compile the blueprint, which caused the collision to break.
@@moviesgamer6574 no I did not. I believe that there is no fix. It's been a problem since UE4. The physic asstes doesn't collide properly with groom and there is no solution, people just don't talk about it. I suggest you look into making hair cards, you might have better luck doing collision with it.
I remember using mm_sd_v15_v2 with BRAV6. I think BRAV6 is compatible with mm_sd_v15_v2 or temporaldiff-v1-animatediff. I think that if you enter (worst quality:1.9) in the negative prompt and enter something like "light gray hoodie", a base person will be generated.
Try adding the following prompt at the end of the prompt. However, please ensure that the total number of tokens is 75 or less. 0: serious face, (look far away:1.3) 4: look far away 5: close eyes, (look at viewer:1.3) 6: (close eyes:1.2), smile 7: smile 8: (smile:1.3) I remember using mm_sd_v14... Sorry, I don't remember...
Sorry, I just don't remember... The material is a 2-second video (16 images) at 8fps generated in SD, and I believe it was made in twice the slow motion.
Wow amazing quality, we tested the early two models a couple of months ago, but found out it needed to use too many iterations to put out great results. We just found out there's a 'mm_sd15_v2' version of the model now. Did You use that newest version or did you manage to get the older ones to run that stable?🤯 In any case, very impressive work!
It is a combination of mm_sd_v15_v2.ckpt and Brav6. 512x720 Steps: 9, Sampler: DPM++ SDE Karras, CFG scale: 6, I use Topaz VideoAI to upscale the size.
If "You are using prompt travel." is not displayed on the WebUI black screen, prompt travel is not enabled. (Terminal window, not web browser screen) If it is displayed, it is enabled. Try keeping the number of prompts (including negatives) below 75.
I don't know because I have never used a hair card at metahuman. I prefer to use groom hair because I can move my hair realistically in the simulation.
I should have come across ur trick sooner, hah . Spend too much time on modo hair rigging for animtion, better get going to UE5 with ur trick. I have just finish her pony tail animation , and sending her to UE5. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-m8lwV3p-llY.html
The top and bottom cameras are captured at the same time in OBS with a 1920x1080 setting. (Create a masked image to show only 960x1080, then mask it in OBS and put the two screens side by side on the left and right). The .mov file to be processed on the command line (PowerShell) must have a timecode attached to it or it will cause an error, Load into DaVinci Resolve and export as 960x1080 top.mov and bot.mov. As soon as DaVinci Resolve starts, I recommend setting the frame rate to the lowest possible 16 fps. Set to 960x1080 and in the edit page, transform "Zoom 2.0 Position X 480 or X-480" to fit a 960x1080 screen. Separate the top and bottom videos. As top.mov and bot.mov respectively. Timeline location should be the same.
Definitely the best hair simulation test I've seen so far. Did you use any Physics plugins to achieve this result? many people prefer the old UE4 physics engine appose to Unreal's new choas physics engine.
I made this hair using Modo 16, which I'm familiar with, because I didn't have enough time to work on it. I used Blender for curve conversion. Although it is possible to achieve the same thing in Blender, some tasks that can be done in just a few clicks in Modo would require node setups in Blender. Therefore, I chose to work in Modo this time. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-_XGZmaTmG8Q.html
