STABLE DIFFUSION 3.5 RELEASED!
Skills:
Image Generation Basics80%
Key Takeaways
This video introduces the new features of Stable Diffusion 3.5
Full Transcript
comfy UI now supports stable Fusion 3.5 just now stability AI released stable Fusion 3.5 stable Fusion 3.5 large large turbo and stable diffusion 3.5 medium these models are free for commercial non-commercial use under the stability a community license y yada uh stability says here 3.5 medium it's not going to be available until October 29th but now comy can support the 3.5 large and the large turbo it seems I think so let's let's check it out so here we have the example workf flows this thing is broken I just stole the devs so let's see if we can get that fixed update to the latest version of comy download the MS models checkpoints well let's let's go do that oh we need to accept this thing ofama jig I agree to the license terms agree freed them there we go ah research there we go all right okay stable Fusion 3.5 large downloads last month six files and versions let's go download uh there we go let's let's just do the large going to put them in models checkpoint and uh there we go now I haven't done anything on SD 3.5 yet so this is brand new to me uh we got some uh human anatomy here we don't have any women lying in grass here do we we have that on this oh here we go so here's a woman lying in grass and it works it works that's cool uh we got to download these so if you have don't have them get them I already have them see if the link is fixed there we go so let's uh let's just do h t fp16 let's try this one we're going to make sure they update our com face I'm just going to go in and do this get cool all right let's start this up where's the workflow uh there it is we don't even have the mods downloaded yet so let's have a look at the workflow here right so sd3 supports different text encoders so it looks very similar to let's see it's a regular load checkpoint Mt sd3 Laten image yeah it's a fairly simple workflow we have a regular K sampler we have a positive and a negative prompt so yeah should be no issu at all we got a little error here but that's should be fine so we're just going to make sure yes we have all of these now we don't have the clip G oh there no oh it should be the same for the sdxl base so we just going to use that if you don't have that you can get it from uh the model manager most likely from the links in the announcement clip name we have that well even have the one named std3 but that should be the same one actually so basically these aren't used right now so all we're going to wait for is this little file up here to finish downloading that should be in about 5 minutes but I'm going to speed this up so you don't have to sit here and wait you can also just quickly check out what they're saying here about the model so we are downloading the stabil fusion 3.5 large one 8 billion parameters so the SD 3.0 medium that we got that was a 2B I think 2B is all you need yeah and this is an 8B so it says superior quality impr adherence and the base model is the most powerful of the stable Fusion family this model is ideal for professional use cases at 1 megapixel resolution so that's about 1024 to 1024 we got the turbo one which is um optimized to work in four steps and the medium at 2.5b to released like we talked about October 29th but I mean we have the large so let's use the large got some uh pretty good examples images here let's see we got some comparisons prompt adherence aesthetic quality uh my favorite here flux Dev currently and we have SD 3.5 large is better in prompt adherence but not as good in aesthetic quality according to this so that remains to be seen so this is kind of cool for three potions the first potion is blue with a label Mana the second potion is red with a with health and the third potion is green with the label poison that does it very well we have this giant turtle or Tor choise T tortoise T tortoise yeah anyway that reminds me of of disc world uh looks kind of cool and free for non-commercial use free for commercial use up to 1 million in annual revenue so that's well most of you watching you are fine ownership of outputs retain ownership of the media generated without restrictive licensing implication so while we continue waiting for the the download there let's check those um so if you have the new comfy which I actually have so I should have used that one uh I did not you can simply import the workflow or drag them into the comfy and we'll Auto download the model so that's really cool uh we can actually try that so let's see if I break anything here now trying to run multiples of these so I recently did a video on this so this is the new comy UI that can run inside of an app and I just want to show you what happens if I were to drop this workflow in without having download anything well I am downloading but not for this installation if that makes any sense we even have some templates uh but I want to do is I'm drop this new one in here ah it's not allowed so I'm sure that's just a temporary thing but uh it's supposed to look like this down so clip clip G safe Chan you have a download here this going to be the sa for any models that um you load into um well that are required by the workflow so that's really really cool but let's head back to our comfy where we have actually downloaded the the model so in this one we have SD 3.5 large and if we cue this up let's see what happens we have a bottle with a rainbow Galaxy inside it on top of a wooden table on a Snowy Mountain Top with the ocean and clouds in the background let's see if this works while I am recording sometimes it crashes when I'm recording but uh we we just going have to see it's kind of cool that we're actually getting zero day support from the release so they're actually releasing it at the same time now I'm not sure why Deion 3.5 had a blog post that was older but to my knowledge this was released now and not a few days ago and here we have our bottle coming in live now now we're running 30 steps CFG 5.5 Oiler schedular sgm uniform this is the default workflow for now and it's actually looking pretty pretty good so let's just try closeup portrait of a Viking woman with battles scars and golden jewelry Nordic wooden F background beautiful scenery so it's try that completely random prompt made up from my mind and what a mind it is so let's see here now here we have it coming in live will we get plastic skin well a little bit so I'm sure you can prompt that better and get some better results but we'll see at least we got stab Shion 3.5 it runs test it let's see if we can get something better than this um not great image the ball that we saw that will look look great test it out for yourself let me know would you think
Original Description
https://stability.ai/news/introducing-stable-diffusion-3-5
https://blog.comfy.org/sd3-5-comfyui/
https://huggingface.co/stabilityai/stable-diffusion-3.5-large
https://huggingface.co/Comfy-Org/stable-diffusion-3.5-fp8/tree/main
Get early access to videos and help me, support me on Patreon https://www.patreon.com/sebastiankamph
Chat with me in our community discord: https://discord.gg/dFB7zuXyFY
Stable Diffusion for Beginners Playlist https://www.youtube.com/playlist?list=PLXS4AwfYDUi5sbsxZmDQWxOQTml9Uqyd2
My Weekly AI Art Challenges https://www.youtube.com/playlist?list=PLXS4AwfYDUi7RvFm4K6lKBH_acaZQMKY4
My Stable diffusion workflow to Perfect Images https://youtu.be/4u-Ytioi3DM
ControlNet tutorial and install guide https://youtu.be/vFZgPyCJflE
Famous Scenes Remade by ControlNet AI https://youtu.be/wVbWZ-Ph9lE
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
Playlist
Uploads from Sebastian Kamph · Sebastian Kamph · 0 of 60
← Previous
Next →
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
How to install stable diffusion tutorial (automatic1111)
Sebastian Kamph
Inpainting in Stable diffusion for beginners.
Sebastian Kamph
OpenAI NEW Whisper is AMAZING!
Sebastian Kamph
Tutorial - Free AI Game assets in Stable diffusion. Episode 1: Sword
Sebastian Kamph
Game assets in Stable diffusion. Ep 2: Jewelry
Sebastian Kamph
Stable diffusion Animation tutorial with AUTOMATIC AUDIO SYNC. Make your own AI music video!
Sebastian Kamph
Stable diffusion img2img tutorial.
Sebastian Kamph
Stable diffusion tutorial - AI Game assets. Episode 3: Treasure chest
Sebastian Kamph
Stable diffusion animation tutorial. Deforum ALL settings explained. Make your own AI video!
Sebastian Kamph
Dreambooth tutorial for stable diffusion. Quick, free and easy!
Sebastian Kamph
Dreambooth to CKPT. NEW VERSION! Dreambooth locally on potato pc.
Sebastian Kamph
Stable diffusion tutorial. ULTIMATE guide - everything you need to know!
Sebastian Kamph
AI music video. Neffex - Winning
Sebastian Kamph
Stable diffusion video input tutorial. How I made this music video singing animation.
Sebastian Kamph
Stable diffusion color grading tutorial. Quick trick!
Sebastian Kamph
Prompt Editing and Alternating Words in Stable Diffusion.
Sebastian Kamph
Stable diffusion gui most important setting. Live render preview.
Sebastian Kamph
NEW Voice2img prototype! This AI assistant is using Stable diffusion!
Sebastian Kamph
Prompts and FREE ONLINE stable diffusion. OpenArt AI tutorial
Sebastian Kamph
Stable diffusion Halloween concept art tutorial.
Sebastian Kamph
Stable diffusion GTA 6 style image tutorial. Quick and EASY!
Sebastian Kamph
Stable diffusion prompt tutorial. NEW PROMPT BOOK released!
Sebastian Kamph
Stable diffusion GTA 6 style image tutorial. Quick and EASY!
Sebastian Kamph
How to install Deforum locally. Stable diffusion animation.
Sebastian Kamph
Dreambooth in Automatic1111. Cpu only & gpu option.
Sebastian Kamph
Nvidia's NEW text to image AI eDiff-I. Will it dethrone Stable diffusion?
Sebastian Kamph
NEW VR in Stable diffusion? The future is now!
Sebastian Kamph
Motion capture workflow implementation with Stable diffusion
Sebastian Kamph
Don't make these 7 mistakes in Stable diffusion.
Sebastian Kamph
Stable diffusion up to 50% faster? I'll show you.
Sebastian Kamph
Stable diffusion 2.0 Released
Sebastian Kamph
Top 5 Stable diffusion tips for newcomers.
Sebastian Kamph
3 AMAZING Stable diffusion models that will change your life!
Sebastian Kamph
Best NEW AI tool? InvokeAI tutorial for Stable diffusion.
Sebastian Kamph
Monetize your AI art on Creative Fabrica with CF Spark.
Sebastian Kamph
NEW Stable diffusion 2.1 RELEASED!
Sebastian Kamph
Stable diffusion 2.1 is GREAT. At this one thing. 2.1 install tutorial.
Sebastian Kamph
Your face in AI images? The EASY way.
Sebastian Kamph
3 FANTASTIC Stable diffusion models you don't know about!
Sebastian Kamph
Unstable diffusion JUST GOT BANNED! 😲
Sebastian Kamph
The end of AI Art? Lawsuit against Stable diffusion
Sebastian Kamph
Stable diffusion TIER LIST. Best GUI ranked.
Sebastian Kamph
Google's ChatGPT rival Bard. Is it better?
Sebastian Kamph
7 Secrets in ChatGPT (Don't tell your boss!)
Sebastian Kamph
How to ChatGPT? Chat GPT explained!
Sebastian Kamph
How to ChatGPT in 20 seconds!
Sebastian Kamph
Midjourney 4C Features
Sebastian Kamph
NEW ControlNet for Stable diffusion RELEASED! THIS IS MIND BLOWING!
Sebastian Kamph
Revealing my Workflow to Perfect AI Images.
Sebastian Kamph
LIVE Pose in Stable Diffusion's ControlNet.
Sebastian Kamph
Control Light in AI Images
Sebastian Kamph
Multi-ControlNet tutorial.
Sebastian Kamph
Control Text in AI Images
Sebastian Kamph
Full AI Art Workflow. ControlNet & Stable diffusion.
Sebastian Kamph
ControlNet Guidance tutorial. Fixing hands?
Sebastian Kamph
Illuminati Model with Noise Offset & Weekly AI Art Challenge
Sebastian Kamph
Paint&Text2Image - MultiDiffusion Region Control.
Sebastian Kamph
Style2Image in ControlNet (T2I)
Sebastian Kamph
Gen-1 AI Animation is WILD
Sebastian Kamph
Famous Scenes Remade by ControlNet AI
Sebastian Kamph
More on: Image Generation Basics
View skill →Related Reads
📰
📰
📰
📰
How Does an Image Model ‘See’? The Weird Phenomenology of CLIP and DALL-E’s Visual Understanding
Medium · ChatGPT
Best AI for Image Generation in 2026: ChatGPT vs Gemini vs Claude Tested
Medium · ChatGPT
Introducing Meta’s Muse Image
Medium · Deep Learning
I stopped writing image prompts and started dropping in reference images instead — here's what I learned
Dev.to · chen mensen
🎓
Tutor Explanation
DeepCamp AI