Finally a FREE alternative to Photoshop GENERATIVE FILL/Expand with opensource tools
Вставка
- Опубліковано 4 лют 2025
- This is amazing! As easy as photoshop but packed with all the opensource power of AI image generation. The future looks amazing. No more credits required to run or play with generative fill.
Hopefully once Affinity get scripting feature like Krita, i would love to see this also in Affinity too.
UPDATE: If you are looking on how to upgrade your krita.ai plugin (to use the FLUX model), check this video
• How to update Krita.Ai...
Link
github.com/Acl...
Thank you for watching and don't forget to hit the like and subscribe buttons.
_____
If you like to support my channel, a ☕ is always welcome. You can use the link below to buy me one. \(^-^)/
www.buymeacoff...
![I Replaced ALL my ADOBE APPS with these [free or cheaper] Alternatives!](/img/n.gif)
Yes, would be nice if Affinity added these generative/removal features.
Affinity is an interface mess. Photopea is much closer to PS in UI though it's greatly inferior to PS (eg. no depth of field blur, can't resize a photo by percentage), but it does have potential. Affinity Photo would have to be redesigned from the get-go. Nobody even tries to compete with Adobe so far.
oh im loving it, this is the most usefull thing i found with sd1.5, and most easiest as a professional
Thanks a lot for reviewing the plugin! It's just fantastic!
The plugin is all the work of one guy called Acly. Been using this for some months especially combining with Blender and mist pass renders to control camera angles.
He is amazing!!!
Krita is an amazing piece of software, and although I am not a big fan of generative AI, i can see scenarios where this could be very useful.
I'll get back to Affinity when they implement this feature.
Great short introduction.
After reading the comments it seems like a lot of people are not installing the image generation part locally but instead using the online option in the settings..
Thank you for sharing.. i am guessing you are right.
Loving the Newark, Nj photo to replace their lamps. ;)
Definitely, looking forward to a more detailed video on live painting.
You got it! Thanks!
I created a file just like you did, but it says "Incompatible document: Color depth must be 8-bit integer (current depth: U16)", even though I created the file with 8-bit depth.
tysm bro, I was getting sick and tired of all those people ripping people off with their "ai generative fill" websites that have some dumb coin system. Like its not even their software bruh yet they are gonna monetize and gatekeep it
Thank you too!
Very cool. Afinity Photo 2.6 will have a great AI selector.
Yes, AP2.6 looks very promising.. hopefully they will add more cool features in par with PS.
Thanks for this demonstration. It was very helpful. Are you going to do an in-depth tutorial on the control layers?
Thank you for comment. That's a great idea.. i'll dig into it. Stay tuned.
@@Graphicalfinity I've been playing with it a lot lately and I love it. Not sure if I like Krita, but the plugin is fantastic! Do you know if it's possible to add more style presets then the 5 standards?
Not only do you need pretty decent GPU, you also need gigabytes of space :D - preferably SSD / NVMe based so you are not waiting for models to load - usually Stable Diffusion based image generator software take easily 20 GB of space with a few models and addons like Control Net etc. (which this seems to be using).
Totally forgot to mention that.. It exactly uses 20GB diskspace :) (without XL and Juggernaut)
having fun on 1050ti 4gb, also 20gb is nothing in todays times, when everyother system has 2+tb hdds and they work fine, no need for nvme just to load a model once
Thanks for the content. Do you happen to know whether it is capable of working with a video, just like in Photoshop?
Not 100% sure but I highly doubt it since Krita is mainly a drawing software
I'm not getting the Fill, Expand, Remove Content, etc. options. I have downloaded and installed the latest plugin. There's no drop-down menu other than 'Generate'
You need to select a portion of the image or the whole image to see those menus
I hope we get this plugin for GIMP.
@purwabrancod also would like it in Photopea
Awesome! Thanks for sharing
Thank you too!
Doesn't seem to work on a Mac and no-one on the github page is able to help
Sadly my mac device is not suited otherwise i would have tried and see if i would be able to help.
It does not work on Mac - you are correct.
It works…
I hope you can help.
I have a question, I want to add an uploaded image instead of asking it to generate an image.
For example , I want to add the image of a handbag I made on a model. Can I upload my image and add it on a generated AI model?
Lastly, you think i can install tgis on a mac laptop?
Thank you for your comment. Regarding your question: If i am not mistaken, you can you just load the image in Krita and then select the area where you want to have the bag and give a prompt to generate it. I am planning on a follow up video and will see if i can show this too in the video. Installing on mac should be possible (according to the docs, you need to make sure you install pyhton and venv first (using the command: apt install python3-venv). Haven't tried it as i don't have a mac with a M chip.
I didnt know free software can be used this easy. Please cover this more indepth
Noted.
Can one use this in Affinity too?
Not yet. I believe was Affinity is working on scripting and when of if they release the scripting feature, it might be possible. Really depends right now on what Canva wants. I can imagine that they might integrate the Canva AI tools (probably with a subscription).
my laptop only have 16gb ram and RTX4050 GPU will that be enough? i dont need it to be to fast just working?
I think that would be good enough. I have 12 MB and it is workable.
Thank you so much for demonstrating this!
My pleasure.
This is AMAZING!!! Winner!
do you have trick to generate image quickly, it takes 3-4 min for 1 generate for me , gtx 1060
I am no expert on this subject, but i think smaller models would definitely (probably less quality) and i would checking ComfyUI optimization tuts (as this utilizes ComfyUI in the backend). At the end it is a lot of experimenting what works, which is not ideal ofcourse. Hope this helps a bit..
my hope is in affinity. come on affinity.
👍😎 hopefully local implementations so we don't get subscriptions. Once they release scripting support, I am pretty sure we'll get tools like this in Affinity too.
i can't find the installer for mac
Beautiful ! 🦋
Thank you! 😊
awesome but i don't find the reference control and i saw it does not like for the moment the XL models. But it's huge anyway love it !
Check the settings and make sure the control nets are turned on. I havent tested with XL model, but it might be that the reference control is not available in XL? Thanks for sharing!
You should mention in the video that it does not work on Mac at all. The required dependencies for PyTorch only exist for ARM macintosh (like M1), and there is some sort of memory consumption issue with M processors where you will not be able to generate locally. What a shame.
Thanks for pointing that out.. It uses comfyUI in the background and if i am not mistaken that should work on the (M) macs without issues...(at least i have seen others running it).
Can I use it with colab/Kaggle GPU ?
When I click Launch at the Server Configure section, I get an error message. "Error during server start up the specific module could not be found." Please help!
goto github and install the missing components if linking to existing comfyui
Yes, more detailed video, please
Noted!
Continuing to make me regret not having NVIDIA / CUDA lol. My first attempt on a 5700xt wasn't great
Thank you for sharing, yep the whole AI scene is based mostly on NVIDIA.
extending images is surely light work?
Why the heck do I need a big graphics card???
I'm disappointed.
I have an integrated graphics card and extending an image should be a simple task for AI.
It is okay if it will work, but just take longer than usual, is that the case?
I think it will work, but will take indeed much longer to generate. The "heavy" graphics card is due to the models being large i think.. even though you need to generate a small piece (like extending), the model still needs to be loaded and processed.. but i am no expert on this matter :)
That's great, but I guess it's not free. When I try to use generative fill it says I don't have enough tokens. Is there any way to use it for free?
Not sure what you are using.. but in Krita with this addon, no tokens required. It all runs on your local machine.
I had the same issue. You have to install a "local server", you do it within the program:
Configure-> Connection-> Local Managed Server. From there, you select a path and click the "install" button.
I thought this is free. It requires 16 token after using it not more than 5 times. Please help
Make sure to go to plugin settings and under connection to use local managed server (and not online service). It might be that you don't have a powerful enough GPU on your machine and maybe this could be the reason it defaulted to online service? (just guessing here...)
I don't get along with the Affinity UI at all. I'm old-school Corel Draw from v 1.0. Affinity's natural family will increase their user base by implementing an effective AI active help system. It really isn't hard. Did this in the early 1990's for quite a few programs..
tu, i love you!! 🐙🐙
Thank you too!
thank you!
Thank you too!
So, where is this plugin pulling its generated content from? Do they have their own database online or using another site?
I would say it's pulling it's data from the models it downloads, just like other SD generators. You can run it locally and offline, so it can't be an online database or website.
Exactly as Elwaves mentioned, it runs a local copy of ComfyUI in the background and once the models are downloaded (probably from civitai or huggingface), no connection is required.
@@Graphicalfinity I don't know why I had any doubt it worked that way but cheers for confirming. 🙂
I installed this last night and after some issues getting rid of my old install, it worked great. Only tested a couple of bits but it's so powerful and easy to use. Thanks for highlighting it.
Yes please!
Thanks
bro thank you so much
My pleasure
Installing it on a Mac is almost as easy as rebuilding the pyramids from scratch. Adobe's generative fill is a censorship nightmare, so I stopped using Photoshop. But the alternative, to be added to the likes of Affinity Photo, has to be easy from the get-go.
Haha.. (wonder if the egyptians from those times would be able to get it installed 😂). If AP would open up an open plugin system, i am pretty sure we would get some cool tools like this in AP too.
@@Graphicalfinity I now it has a new owner, the the point of AP has always seemed to make it as troublesome to use as possible. It's incredible that Photoshop has no good competition. It has competition for basic stuff, but no other app really lays a glove on it for full-featured editing. No, I have no faith in Apple. They will either do a half-assed job or never get out of the gate. Apple's capacity for useful innovation ran out of gas years ago. Granted, there are goggles, which have as much potential as 3D TV.
The generator fill its not free
Yes it is... just generated content without using any account and paying anything.
More detailed please
Noted!
Not free needs a token when you log in your account.
There is no need to login to anything to use it.. Don't understand what you are referring too.
gg man love ya
My pleasure.
I think you have just single handedly made me cancel by adobe subscription with this
:)
@@Graphicalfinity UPDATE : I have now finally cancelled it after years of being subscribed. I LOVE this Krita plugin. Thank you buddy!
@@TPCDAZ how does it compare?
WOW
this is biggg
too slow! photoshop does it in half a second
Its free :) also it depends on your video card
I hate these generic AI „illustrations“ but go ahead with that mindless sh…it
I totally agree with you regarding art/illustration/photo generation, but the AI tooling for photo editing, like auto subject selection, tracing, removing/replacing content etc is a big plus as these save a lot of time in daily use.
Maybe many years ago, some photographers/artists thought the same about Photoshop and some found it useful, i think this is the same with AI eventually. It is not the fault of the tool but the ones using it.
@Graphicalfinity Agreed, sorry for my bitterness. You're a cool guy, I see a lot of potential for indie creators, I also see the damage it's causing a lot of designers ending careers, I stand by how I feel which is for AI generative software to work it needs to still allow complete user customization of their projects.