วีดีโอ

thank you for the easy to follow video. A lot of great advice

Hey Rob, what was the node you use for exporting out the entire workflow as a .png screen grab?...It's really helpful but I can't for the life of me remember. Thanks

very good workflow but unfortunately it stopped working after the last update Comfy gives an error Loop (67,576,575,574,....) - not submitting workflow. can you advise how this could be fixed?

superb!

Great wf! One idea: could it be that instead of rewriting the prompt you could use Florence to rewrite the prompt for that specific refining area? I usually try to do that with a extra text concatenation if necessary. Thanks!

Hi there, and thanks for all the awesome content! I'm super grateful for all the knowledge you're sharing and I love your generations, they are impressive! Do you have a civitai account? Or a place you post your final images? And 2nd, since you use rgthree nodes (you may already know this - sorry!) a tiny QOL improvement for group mute/bypass toggles: 1. go to setting 2. click on 'rgthree', then click on 'rgthree-comfy settings' 3. scroll down & enable 'Show fast toggles in group header' with 'mute and bypass' & 'always' (also has option for only mute or bypass) 4. enjoy one click bypass for groups (at least at group header level, situational I guess, but I liked to sometimes get away with just one click - I know there's also the fast groups node - I feel this completes the set of options available for muting/bypassing)

No JSON ?

Lovely workflow. Thanks sir!!!

Actually it understands prompting better than any model before, so its easier. The problematic part is that the dev model is so huge that no consumer class gpus can handle it before 5090 coming on the next christmas.

This is great! So glad you are sharing your insights and along with the workflow, I appreciate thank you.

Wow really nice workflow thanks for your work. Honestly, just add Segmentanything2, ic relight, and controlnets options and it would be just perfect for almost any tasks. Also, can you upload the .json file of the workflow please, here on the drive it's just a png of the workflow =D

I've noticed that Flux can be stubborn with image to image too. It doesn't like to change things a lot, which might be good for photographic resampling/upscaling where you want details preserved as much as possible, but it's bad for creative resampling/upscaling. Has anyone been able, for example, to transform a cartoon image/painting into a photo or vice versa? I've not been able to do this with Flux so far. It's a lot easier with SDXL and SD 1.5.

Always pleased to see when you've uploaded a new video Rob

I'm having a hard time understanding most of this. Showing this to my grandmother would certainly give her an aneurysm.

Nice work. Thanks for the detailed explanation.

I really appreciate this kind of video, it saves everyone a lot time but also opens up new lines of enquiry. You could go even more granular with these. There's definately some weird weighting in the model. I'd love more like this, I'm trying to find photographic styles and oddities. Who knows, maybe we can find some magic in latent.

I'm impressed with Flux's ability to understand plain English prompts rather than a set of keywords, also impressed that it works on my 4GB VRAM card (5 to 20 minutes per image), but I'm playing with SD1.5 models that produce better images... mostly.

What a fun project to do. Thanks for going through the pain of putting it together so we didn't have to.😄. It was really interesting.

Yup. Flux terrible in styles or Artists. Waiting loras for that.

interesting, but the thing most of us probably need is just how to get started, I mean just loading the simplest image was impossible for me, and I ran out of my 15 minutes free time before I had even done anything.

Hmm, looks like at least right now Flux is mostly good for pretty women and food. Latent vision was saying it did better with more verbose sentences rather than keywords. Like it was trained on generated image descriptions.

nice workflow! maybe can you also create one to kitbash a image, e.g. create a baseball cap with logo and sticker (logo and sticker from separate images)

Simpletuner now works with FLUX for fine tuning! 😊

Thanks for the video!

Good feedback. Thanks! Btw you have upgraded your mac ram? Of get a new rtx 4090 ? 😅

Thank you very much, Rob, for your honest review. I haven't started with FLUX, but I have seen some vids where the basic wokflow was made in SDXL (w. Controlnet, Loras etc.); the refining was done with FLUX. I'll start it when there is more light about cfg, guider, noise and some working Controlnets/Loras are available. Nevertheless always fun to listen to your explanations and detailed findings.

perfect timing for this video :)

Grandiose!

Thank you for this flow! It does not seem to work well for cartoony/more stylised sketches though :/

What does styleske represent?

This is perfect Rob, been learning Alpaca, thanks!

Really Great Workflow and walkthrough - thanks a bunch!

can you provide the json file?

Darn, the workflow you link is different than the one you use in the video and I can't get it to work the same.

I've been putting off learning how to do workflows and templates properly. This video gave me all the answers--cheerful and charming as always!

it worked well for me. thank you for sharing

Super fun! Amazing results! Thank you for sharing the workflow.

Ive wanted something like this forever thank you so much - we need to be doing more with the latent process

Thank you for the great videos Rob! Keep em coming! Can you please take Kwai Kolors for a spin it looks like and amazing model.

is there any clean way to up the batch count on this workflow, im noob sry if its a dumb qeustion

Thanks Rob! Your tutorials are always helpful. Cheers!

Hello, I do everything as in your video, but I get very grainy fragments of images if I divide the CLIP into zones, neither changing the resolution nor increasing the resolution helps me. If I make a canvas for 2-3 CLIP zones, then these zones are grainy and of low quality. I would send you a picture, but I can't leave a link here... I found a solution to this problem, it is to connect a whole CLIP that will cover the entire canvas, but I think this is wrong, it distorts what is desired, but the elements do not lose quality. Can you answer me here, please?

Is there any tutorial to teach people how to add controlnet nodes one by one and not to download someone's workflow?

Thank you for sharing.

Great explanation of doing additional steps to improve the image quality! Thank you very much, Rob! The idea of painting parts of the mask area into the image is really great and opens the door for solving some hand problems as well. By the way, are you using a grid tool for your crop solutions cutting out parts of the image?

I like your ComfyUI info, but man you really need to have better audio.

Hey Rob, Thanks for posting your exploration of ComfyUI. I've spent a fair bit of time exploring the intricacies of prompting but am pretty far behind on more advanced workflows. I enjoyed your video some months ago on how closely aligned different artists styles are within Stable Diffusion and am now subscribed.

There is a lora, EnvyCelShadedXL01.safetnsors, referenced in the JSON you provided. I haven't been able to locate this using google or in civitai. Where might someone find this lora?

Do you think one or several IPAdapters could work instead of the LoRAs?

This is priceless!