Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: New AI edits images based on text instructions

github.com

Post reply on HN
Print view View original

211–220 of 241 posts

Re: Show HN: New AI edits images based on text instructions

#211

Post by goffi »

Wow that's really impressive (I've seen similar things in research papers for a while now, but having it usable so easily and generic is great).

A few questions:

- would it be possible to use this tool to make automatic mask for editing in something like GIMP (for instance, if I want to automatically mask the hair)?

- would it be possible to have a REPL or something else to make several prompt on the same image? Loading the model takes time, and it would be great to be able to just do it once.

- how about a small GUI or webui to have the preview immediately? Maybe it's not the goal of this project and using `instruct-pix2pix` directly with its webui is more appropriate?

Thanks for the work (including upstream people for the research paper and pix2pix), and for sharing.

Re: Show HN: New AI edits images based on text instructions

#212

Post by karim79 »

throwaway675309 wrote: post #194
karim79 wrote: post #190

I've been toying with SD for a while, and I do want to make a nice and clean business out of it. It's more of a side-projecty thing so to speak. Our "cluster" is running on a ASUS ROG 2080Ti external GPU in the razer core-x housing, and that actually works just fine in my flat. We went through several iterations of how this could work at scale. The initial premise was basically the google homepage, but for images. Th…

A ticketing scheduler system is how 99% of systems that require long running CPU/GPU intensive that cannot be run in parallel are implemented. It's how I built up my stable diffusion discord bot which is backed by a single RTX 2060. I'm glad you have this working but I wouldn't exactly call this "solving the scaling problem", you're just running it in a blocking "serial fashion". With enough concurrent users it could…

Agreed. The scaling model/queueing system was implemented as a POC which can be scaled by plugging in more cards/hosts. I hope we can find the time to animate this soon enough.

Re: Show HN: New AI edits images based on text instructions

#213

Post by bryced »

goffi wrote: post #211

Wow that's really impressive (I've seen similar things in research papers for a while now, but having it usable so easily and generic is great). A few questions: - would it be possible to use this tool to make automatic mask for editing in something like GIMP (for instance, if I want to automatically mask the hair)? - would it be possible to have a REPL or something else to make several prompt on the same image? Load…

> would it be possible to use this tool to make automatic mask for editing in something like GIMP

probably but GIMP plugins are not something I've looked into

> REPL

already done. just type `aimg` and you're good to go

> GUI

GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews?

Re: Show HN: New AI edits images based on text instructions

#214

Post by iuiz »

bryced wrote: post #84

Here is a colab you can try it in. It crashed for me the first time but worked the second time. https://colab.research.google.com/drive/1rOvQNs0Cmn_yU1bKWjC...

I could not get the first cell to run.

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. tensorflow 2.9.2 requires protobuf=3.9.2, but you have protobuf 3.20.3 which is incompatible. tensorboard 2.9.1 requires protobuf=3.9.2, but you have protobuf 3.20.3 which is incompatible.

Re: Show HN: New AI edits images based on text instructions

#215

Post by sseagull »

bryced wrote: post #213
goffi wrote: post #211

Wow that's really impressive (I've seen similar things in research papers for a while now, but having it usable so easily and generic is great). A few questions: - would it be possible to use this tool to make automatic mask for editing in something like GIMP (for instance, if I want to automatically mask the hair)? - would it be possible to have a REPL or something else to make several prompt on the same image? Load…

> would it be possible to use this tool to make automatic mask for editing in something like GIMP probably but GIMP plugins are not something I've looked into > REPL already done. just type `aimg` and you're good to go > GUI GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews?

> GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews?

Somewhat OT, but I find this really funny. It says a lot about the difficulty of using various ecosystems and where communities spend time polishing things.

"Yeah, I made something that takes natural language and can do things like change seasons in an image. But a GUI? That's complicated!"

It's not a criticism of you, but the different ecosystems and what programmers like to focus on nowadays.

Re: Show HN: New AI edits images based on text instructions

#216

Post by kleiba »

wongarsu wrote: post #173
kleiba wrote: post #164

Earlier quoted context omitted.

> + transmitted over internet (ie state lines) But OP specifically said: "the tool is executing on my laptop. - the output image is not shared with anyone."

Maybe use a desktop instead, because crossing state lines with that laptop might carry a jail sentence with a minimum term of 5 years.

I would assume that running an AI image generation software is something you'd do on a desktop anyway.

Re: Show HN: New AI edits images based on text instructions

#217

Post by bryced »

sseagull wrote: post #215
bryced wrote: post #213

Earlier quoted context omitted.

> would it be possible to use this tool to make automatic mask for editing in something like GIMP probably but GIMP plugins are not something I've looked into > REPL already done. just type `aimg` and you're good to go > GUI GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews?

> GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews? Somewhat OT, but I find this really funny. It says a lot about the difficulty of using various ecosystems and where communities spend time polishing things. "Yeah, I made something that takes natural language and can do things like change seasons in an image. But a GUI? That's complicated!" It's not a criticism of you, but the dif…

Fair but I'd point out I also didn't make the algorithm that changes photos. I'm wrapping a bunch of algorithms that other people made in a way that makes them easy to use.

It's not just that GUI's are hard, it's that the "customer" base will inevitably be much less technical and I'd receive a lot more difficult to resolve bug reports. So no-gui is also a way of staying focused on more interesting parts of the project.

Re: Show HN: New AI edits images based on text instructions

#218

Post by bryced »

iuiz wrote: post #214
bryced wrote: post #84

Here is a colab you can try it in. It crashed for me the first time but worked the second time. https://colab.research.google.com/drive/1rOvQNs0Cmn_yU1bKWjC...

I could not get the first cell to run. ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. tensorflow 2.9.2 requires protobuf =3.9.2, but you have protobuf 3.20.3 which is incompatible. tensorboard 2.9.1 requires protobuf =3.9.2, but you have protobuf 3.20.3 which is incompatible.

I believe you can just ignore that error. The cell ran.

Re: Show HN: New AI edits images based on text instructions

#219

Post by sandworm101 »

kleiba wrote: post #164
sandworm101 wrote: post #158

Earlier quoted context omitted.

Nope. It is illegal if you do it on ms paint too. Images of naked kids (real or fiction) + transmitted over internet (ie state lines) = bad bad day for all involved.

> + transmitted over internet (ie state lines) But OP specifically said: "the tool is executing on my laptop. - the output image is not shared with anyone."

The state lines thing is only about federal crimes. Possession is most likely illegal as a state crime on its own. The repercussions of even being suspected of crimes in this area means that every rational person would take precautions far above what is legally necessary to prevent any association with such material.

Re: Show HN: New AI edits images based on text instructions

#220

Post by goffi »

bryced wrote: post #213
goffi wrote: post #211

Wow that's really impressive (I've seen similar things in research papers for a while now, but having it usable so easily and generic is great). A few questions: - would it be possible to use this tool to make automatic mask for editing in something like GIMP (for instance, if I want to automatically mask the hair)? - would it be possible to have a REPL or something else to make several prompt on the same image? Load…

> would it be possible to use this tool to make automatic mask for editing in something like GIMP probably but GIMP plugins are not something I've looked into > REPL already done. just type `aimg` and you're good to go > GUI GUIs add a lot of complexity. Can your file manager do thumbnails and quick previews?

thanks for the quick answer and cool for REPL. Yeah sure I can just launch Gwenview on the output directory.

> probably but GIMP plugins are not something I've looked into

I was just thinking about a black and white or grey level output image with the desired area , no need to integrate it in GIMP of whatever. I've tried a prompt like "keep only the face", but no luck so far.

Post reply on HN