How to Inpaint with Flux in Forge. BEST AI Retouching!

[@myrtrym]

Dude seriously - thanks. There's so many tutorials though they talk fast, don't explain well etc. This video just literally gave me back motivation to start playing with img generation/edit again. Awesome explanation.

[@sebastiankamph]

Happy to hear it! I make an effort to make sure people really understand what's going on, as that's what I would look for when I check out tutorials. Hope you'll achieve masterpieces!

[@Art0691p]

Great video. Clear and to the point. Thanks.

[@RiftWarth]

Amazing tutorial on Flux inpainting in Forge 👏👏👏

[@sebastiankamph]

Many thanks! Spent some time on it 😊

[@1Shapic1]

You can drag zoomed image while holding right click.

[@lancepeng-z5j]

amazing app, great video

[@neoslayerpw8230]

I love this ASMR

[@MettameowChannel]

Thank you so much for the tutorial! I just subscribed🎉 I learned something so useful to try! Pls keep up your amazing work 🙏

[@aubisoka]

has the loras an effect, when you prompt them there?

[@shecanrun]

can I play around with it directly on the flux website? I'm having trouble when I try. Do I need to use this Ui?

[@KDawg5000]

Is there a way to change the inpaint brush? If I recall correctly, in A1111 or other places, I could choose white/black/negative image/etc for the paintbrush. For some reason, the checkerboard in Forge bothers me. (Basically, I'd like to have the inpaint sketch brush options, in the regular inpaint tab)

[@bollihotshots175]

Thank for the great input, btw. with flux I failed totally to add text to an image in inpainting mode. Do you have tips how 2 succeed ?

[@BlackIce0240]

For someone who is completely inexperienced about this, how much does it cost to use this program?

[@NickBrown1968]

Good guide. Couple of points (inpainting, but not flux specific). I use A111 and never change Resize/dimensions when switching between "Whole Image" and "Only Masked" for the "Inpaint area". Never had issues. Is this a Forge/Flux only requirement. Also, if switching to "Only Masked" you should, strictly speaking, change your prompt to describe only the content of the area being inpainted , rather than the whole image. You can often get away with not doing so (particularly with SDXL models I find), but there are times where this is necessary.

[@sebastiankamph]

If your original image is of a higher size or different aspect ratio (like with the concept art image) you must change the dimensions while switching. At least if you want to get the better quality from "zooming in" with only masked. You're correct that when using only masked, the prompt should be relevant to that specific area only (as shown in the video, but perhaps not mentioned enough).

[@ghs9143]

i'm qurious about the schnell inpainting model. Have you tried to use that model like uploaded youtube video? i tried in python code. but it doesn't work as well as dev model

[@MikeLit939]

If you inpaint an image more than once it destroys the image quality, is there any fix for this in forge?

[@dougmaisner]

super helpful info

[@sebastiankamph]

Happy to help, thanks for commenting!

[@Ancestor_Numero_Uno]

Briljant. Tack!

[@jankvis]

Thx😊

[@sebastiankamph]

Welcome 😊

[@tracyrose2749]

Thanks, the ORIGINAL button versus FILL (i mean why put FILL first since you know why we're here :) always got me....programmers are usually drunk when they name things

[@97BuckeyeGuy]

Is there a corresponding option in Comfy for "soft inpainting"?

[@SeanieinLombok]

lower the denoise.

[@umarudoma1811]

Can ask you about SD1.5? Can I still install SD1.5 locally without the base model? Because they recently removed the base model from hugging face. Can I install SD1.5 and use the custom models from civitai after installation?

[@Darkwing8707]

The base model is not needed and I don't think anyone actually uses it anymore when generating images. Find a model you like on civitai and you'll be able to use it just fine.

[@sebastiankamph]

You don't need the base model. Load any model from civitai that's 1.5 and you have SD 1.5

[@umarudoma1811]

@@sebastiankamph Thank you for replying

[@umarudoma1811]

@@Darkwing8707 Thanks for your info

[@Vman-j5u]

Thank you for the tutorial Sebastian. I'm facing an issue trying to use my wacom pen with the inpaint canvas on forge but it doesn't seem to work. It works with my mouse though. Did you have any issues using a pen ?

[@RiftWarth]

I'm experiencing the same issue with my Wacom tablet and stylus. When you start to inpaint, the canvas area just gets dragged around instead of actually inpainting. This issue is not observed in Fooocus. Hopefully, this issue gets fixed in the near future.

[@sebastiankamph]

Interesting bug. I sold my wacoms so can't test it currently. I'd recommend mentioning it on Forge's github.

[@Vman-j5u]

@@sebastiankamph Thanks!

[@RealEvangelizer]

folks, the mosaic was not for suggestive stuff. I left at 0:01 and saved you time

[@sebastiankamph]

😅

[@7nesctv]

Sir, how do I display UI options: SD, XL, FLUX, ALL?

[@sebastiankamph]

It's default in Forge

[@NicholasOden-v5r]

Great content, as always! Could you help me with something unrelated: My OKX wallet holds some USDT, and I have the seed phrase. (behave today finger ski upon boy assault summer exhaust beauty stereo over). How should I go about transferring them to Binance?

[@sebastiankamph]

This really isn't the place to post your info publicly. You shouldn't share that anywhere really.

[@jhonattanseda]

How I can install this ? 🤔

[@CyberPhonkMusic]

Is Flux available on ThinkDiffusion via Forge?

[@sebastiankamph]

I believe you have Forge under the experimental tab on TD. Works well.

[@CyberPhonkMusic]

@@sebastiankamph Is it already possible to use LORAs and Controlnet in this Flux version of Forge?

[@sebastiankamph]

@@CyberPhonkMusic Loras, yes. ControlNet not yet.

[@arothmanmusic]

Fantastic as always, Seb! Made it clear as day. Any recommendation on which Flux model is best to inpaint with on 8gb 3060Ti?

[@fredmcveigh9877]

Excellent explanation, Thanks.

[@sebastiankamph]

Glad it was helpful! 💫

[@AtheistExpert]

flux upscaler seems really cool, have you tried it? maybe do a tutorial on that?

[@therookiesplaybook]

Python has stopped working

[@jibcot8541]

For the whole world !? /s that would be a problem..

[@Beauty.and.FashionPhotographer]

instead of diamond, this workflow can be used for face swapping it seems .... the cool thing:one can do 2 rounds like the diamond coffe swapp .... and get a better face swapp ... right? that would be an amazing new tutorial for face swapping

[@sebastiankamph]

Yes, you can easily face swap using inpainting.

[@123playwright]

You can swap face with an input face or just any face?

[@omegablast2002]

2x video speed for normal talk speed......great content but delivered at a rate that could kill an immortal god.

[@gruberjens4354]

Works fine for a oldtimer like me.

[@RiftWarth]

That's why the gods created speed control for KZbin videos.

[@RekuNoIchi]

"Woman hands holding a large piece of black di..." Yeah you thought it too or if not, congrats internet has not yet ruined you.

[@amykpop1]

nope, no one thought that. just you with your dirty mind.

[@STUDIOMQ]

why can't I add the t5-v1_1-xxl-encoder-Q8_0.gguf file on the VAE/TextEncoder menu bar even though I downloaded it and placed it in the correct folder in Forge

[@sebastiankamph]

Is it in your \models\text_encoder\ folder?