Deforum Settings Explained - Part 1 Stable Diffusion Automatic 1111
ฝัง
- เผยแพร่เมื่อ 27 ก.ค. 2024
- In this Video I will explain the Deforum Settings for Video Rendering with Stable Diffusion. We will look at the Render Settings, Sampling, Resolution, Seed, Sampler, Steps, Batch, Batch-name and Seed Behavior.
We will explore how to write a promt for Deforum and how to start Prompt at different times in video with the Frame number.
In the Animation settings, we will look at the Motion parameters for 2D and 3D settings. Angle, Zoom, translation x y z, rotation 3d x y z. Also the Generation settings for noise, strengh, contrast, cfg scale and may more setting.
Install Automatic 1111: • Stable Diffusion Insta...
Install Deforum for Automatic 1111: • FREE Video AI - Deforu...
Support my Channel:
/ @oliviosarikas
Subscribe to my Newsletter for FREE: My Newsletter: oliviotutorials.podia.com/new...
How to get started with Midjourney: • Midjourney AI - FIRST ...
Midjourney Settings explained: • Midjourney Settings Ex...
Best Midjourney Resources: • 😍 Midjourney BEST Reso...
Make better Midjourney Prompts: • Make BETTER Prompts - ...
My Facebook PHOTOGRAPHY group: / oliviotutorials.superfan
My Affinity Photo Creative Packs: gumroad.com/sarikasat
My Patreon Page: / sarikas
All my Social Media Accounts: linktr.ee/oliviotutorials - แนวปฏิบัติและการใช้ชีวิต
Support my Channel: th-cam.com/channels/CKx8mAHiFus-XYQLy_WnaA.htmljoin
Install Automatic 1111: th-cam.com/video/Pyze0seDHzA/w-d-xo.html
Install Deforum for Automatic 1111: th-cam.com/video/Zvo86Y7hgfY/w-d-xo.html
Idk if you could help me but when i run the webui-user tells me "Couldnt launch python exit code 9009" but i do have installed Phyton so...
definitely my favorite ai tutorial channel. clear breakdowns. would love to see more examples along with each setting. and maybe time stamps for future. thank you!
Your videos helped me to set this up locally and wrap my head around the settings. The animations this can produce are simply mind-blowing, I now have an old PC creating animations constantly many thanks!
do you have nvidia?
Y-Achses spricht man wie das Wort "why", nicht "epsilon"
Und "copy" spricht man "koppi" und nicht "kohpie"
Ich liebe ich deine Videos und bin dir sehr dankbar für deine Tutorials! Aber das musste ich einfach mal los werden ;)
Thanks so much Olivio! I was able to get everything set up and working because of your tutorials - can't wait to follow along with this! 🙌🏾🙌🏾
Awesome, thank you very much :)
Amazing explanation and breakdown. This answered so many of my questions! Thank you!
Thank you so much. Looking forward to part 2.
Astonishingly helpful. Thanks!
nice explanation , i m so hype for the part 2 !!!!
Very helpful! Thank you for all your work! 🎉
Fantastic work, I've been too intimated by this for a while now, but this really helped break it down! Thanks :3
Awesome, you are welcome :)
Excellent content as always, exactly what I was waiting for before going all in and fumbling around with these settings haha.
Thank you :)
Thank you so much Olivio:) ! this was really helpful!!!,,, still having trouble getting my prompt to change from 1 key to another, but still,, this is really really great information:) !
Thank you Brother, you've helped me a lot!
fantastic explanations!
Good Job Olivio ! Hope the " input video " will be functionnal soon !
Thank you for the information.
Great video! Did you ever release the part 2!?!?
Great video -- Part 2?
Hi love the video. In part 2 it would be awesome if you could show how to morph from one predetermined seed to another. I still haven't been able to figure that out. I want to be able to guide a video through images that I perfected in text to image then move a video through those images. I feel I am so close but just can't get it. Also if you happened to record a how long form how to on how you made your awesome music video ot would get a big thumbs up from me for sure. Thansk for all the info you provide!
I would like to know this as well .... seed progression
great video as allways. Tough for me I cant understand the hype of Deforum
For cadence = 2, the frame starts from 0. Therefore, 0 frame is created then 2 frames. And 1 transition frame is created between. But be careful of your frame per second. For 30fps, every second 15 frames are created. For 24fps, only 12 frames. For 12fps, only 6 frames. Once you understand that, you can set your cadence accordingly to your fps and get a smoother and coherent images.
Thank you for this information. based on what you're saying, if im using 15fps at cadence 1, what fps would i have to set to get 15fps at cadence 2?
hey Olivio, wish all is well and thanx for the awesome videos!
QueStion is the pruned-emaonly Model the same as the collab notebook version 0.5?
So can I generate the same results with the notebook v 0.5 (if using same settings and seed from auto 1111)?
Could you please do a video on getting SD working on an AMD graphics card on windows? I feel you explain things in a really easy to understand format for us non coding plebs and I'm struggling here.
Thanks for all your good content thus far!
add 1.thanks
When do we get part 2 where you go through Init and Video Output? I'm stuck
Me too bro
The day when you change your background to dark, I'll come back again!
Have there been significant changes with Deforum since this video released?
Trenning on artstation - trending was wrong. Thanks for the explanation video :)
Thank you, yep that's my dyslexia for you ;) in most cases the AI understands what i mean regardless. But you are right, i need to pay more attention to the prompts
In the latest update of Deforum for Automatic1111 the image is not getting blurry if you zoom. Looks way better. I guess it's a version 0.5 beta.
Thank you so much for your videos . Please let me know
Where can i get different models as you have ??
Were you able to get Deforum 0.6 working with automatic 1111?
i'm trying it with arcane v3 models and using lyrics songs as prompt to see what happens so far it works step by step. i just have to make a video editing to see the final result
Ohh that’s interesting.. do post the results!
Easy is always best for me since I have level 0 programming skill.
👍
no coding skills needed :)
thank you for your always clear tutorials.
I have a question:
I use Mac and when I generate FFMPEG video it says "Error: FFmpeg not found. Please make sure you ha e a working ffmpeg path”
Why?
Near and Far settings refer to the clipping plane. Near and far clipping planes are imaginary planes located at two particular distances from the camera along the camera's sight line. They determine how much of a scene is seen by the camera in the viewport. Only objects between a camera's two clipping planes are rendered in that camera's view.
can you please simplify this a little bit more with an example because I am a visual learner and I'm having a hard time visualizing what you're saying and how the values impact that
thanks for the useful video, when i put ZOOM 0:(1) the image starts to go into somethink wireframed , what i am doing wrong ?
Which setting applies to the strength of the prompt used when influencing the input? I'm using an uploaded video and just wanted to apply my prompt by ~25-50%, flavouring the video instead of completely reworking it
Hi, just installed deforum into SD and I am wondering if there is a way to upload personal videos. When doing via colab there is an upload video given but, not in the local install.
Is there any description of how to use the math functions in the prompts? My results do not look like what I would expect, if it is working.
Hi! I’m a very beginner with this and the reason I started to follow tutorials for Stable Diffusion is because I would like to learn to creat this king of videos for the Fencing club I’m helping with some digital content. So my question is if I need some special requirements for the installation? Can I do this on my laptop?
Hi, where do I specify path to input sequence images or the video for video mode?
Near and Far is for camera clipping. How close something could be to the camera before it disappear and how far camera could see in that virtual space
bump
so if you're doing a video where it's zooming in and you want the subject to take up let's say 25% of the screen and by the keyframe when it changes to something else you want the subject to be let's say 75% of the screen/frame how would you set those numbers vs very far away like 10% of the scene to 25% of the scene? would FOV also come into play?
How do you load the custom settings from the text file to place it in the custom settings spot? I tried copying the complete language and pasting it there, and tried copying the location of the file and pasting that there, but neither method worked.
Can you please explain how do we make with google colab? Btw I want to u huge thx for every videos!
do you know how to increase the maximum number of sampling steps in deforum for automatic 1111?
You have f111 installed, you naughty boy.
BTW, f222 is out.
only for research reasons of course ;) Also I think you get better hands and more anatomically correct people with that, even if they wear cloths. Thank you for letting me know about f222
@@OlivioSarikas 👍👍👍
how many you can add for negativ promots? by coma as much as I want?
Is there any way to do strength scheduling in 1111?
I followed your other video you mentioned to install this to the T, however my Deforum tab has about half of the settings you show here. Did I do something wrong?
they changed the UI design.
So Deforum is basically a camera animation tool. Is it possible to selectively animate the individual elements within the content composition? Perhaps using depth maps or other ways?
There are alpha masks., but you can't have different prompts for different areas of the same video scene. However, you can of course render scenes and then combine them later in a video cutting software. Or you can combing video materials before hand and then use that video input as a source to be rendered on
very good videos you have. thanks for that. i was trying to make video/gif, just with the default setting... but where does it save the images/video/gif? i changed the file path the secod try... but found no images at all? it did do the animation. you can view it in the "stable diffusion"... but where are the files? thanks. :D
sorry... found them. i need to buy new glasses... sorry. :D🤣
+1
Hello, how do I upload photos already taken to Deforum?
I'm missing a lot of options, like the Highres fix, seed_behaviour, etc. Does anyone know why this is? Were they removed in later versions or am I using a wrong version myself?
did you find out why it's missing?
Error: ''(' was never closed (, line 1)'. Check your schedules/ init values please. Also make sure you don't have a backwards slash in any of your PATHs - use / instead of \. Full error message is in your terminal/ cli.
what to do?
Any reason you're not using safetensors?
PLease Help, how to load settings from file? This version is different
I have installed the deforum, but it is not appearing on the tab button. Is there any way I could resolve this?
hey guys.. i want to add an Image Init to Deforum in A1111 but i get always this error:
Error: 'Init image path or mask image path is not valid'. Check your schedules/ init values please. Also make sure you don't have a backwards slash in any of your PATHs - use / instead of \. Full error message is in your terminal/ cli.
Image path is: C:/webui/images/01.png
i tried it with backslash, upload the image on the web and use a web link, tried another image, another path.. nothing work :(
Is there any way to run these with GPU having 8 GB or less Vram ?? ☹
Hi, can you help me with that issues? User friendly error message:
Error: Init image path or mask image path is not valid. Check your schedules/ init values please. Also make sure you don't have a backwards slash in any of your PATHs - use / instead of \.
see if there some values in kernel, are bad .
can i use lora in deforum?
I am getting blurry videos, first frame load great, then all brury .. do you know what is wrong?
Is there a "Part 2"??
Having problem with this ERROR: List index out of range - Someone know how to fix?
Hi,about the animation mode, I can only use 2D, if I click 3D or video input, then comes error ：FileNotFoundError: [WinError 2].
Why? plz help!
yes, there are 2 files you need to download. I will make a video about it
thanks, this helped a lot.
tho im still running into problems, i'll try and find a solution online
it generates all the images for the video, but when it finishes, it fails making it into a video with a file not found error. im not sure if im the only one having this problem
set it to gif .. ive never had a problem - even if I interrupt the render
How do you start with a seed image?
How do we transition from a specific seed to another specific seed?
i would think that in the seed schedule you can set the starting points for different seeds and it should transition into that seed inbetween, but i will check on that
@@OlivioSarikas any news on the seed schedule ?
When is Part 2 coming? I can't seem to find it!
did you find it, i cant either
can you do it with google colab??
Getting a weird Python error after using it. Was working, restarted it and now it's not.
I can't find part 2
Can you please explain me where do i find the video outpui? Strangely enough i cannot find any file :/ Thanks a ton
did u figure it out
@@rubaduba8263 - the file is in the the 'img2img' folder, under the 'output' folder in the 'stable diffusion' directory. mine is D:\AI\stable-diffusion-webui\outputs\img2img-images\ animationame
@@MichaelFlynn0 tell him that not me
the file is in the the 'img2img' folder, under the 'output' folder in the 'stable diffusion' directory. mine is D:\AI\stable-diffusion-webui\outputs\img2img-images\ animationame
@@rubaduba8263 - woops
playback speed set to 1.25
works great with the rule34 model incase anyone wanted to know
Where's part2?
The video is unsharp:-( Can not read it!!
Error completing request
Arguments: ('a beautiful, woman-maid, wearing body armor made out of vacuum cleaner parts, determined to win, post apocolyptic living room, ultra detailed, full shot, ultra realistic, keyshot -ar 16:9', '', 'None', 'None', 20, 0, False, False, 10, 1, 7, -1.0, -1.0, 0, 0, 0, False, 512, 512, False, 0.7, 0, 0, 0, False, False, False, '', 1, '', 0, '', True, False, False) {}
Traceback (most recent call last):
File "C:\Users\gabda\Documents\SDlocal\modules\ui.py", line 185, in f
res = list(func(*args, **kwargs))
i got some of this errors, what i need to do?
It takes 30 minutes to create a 4 second video. Sad
You need a better GPU or a online service. It's really fast with good gear