The DIY forge - AI content Tutorials and Coding

AI-based Video to video generative tools (vid2vid)

Details: Written by: Super User; Category: vid2vid; Published: July 19, 2023; Hits: 1457

Do you want to convert a video into another video ?
For example, do you want to convert a real video to an anime, or replace the Ferrari that you use on your video with a Toyota Prius to impress your tax collector ?
With the latest development of artificial intelligence, this becomes feasible.
Here is a small collection of tools:

Convert an anime character to a real one [youtube link]

Details: Written by: Super User; Category: image to image; Published: July 19, 2023; Hits: 1379

Here is a link to a tutorial on how to convert an anime character to a real one:

Click on the thumbnail above to watch the youtube video.

Some news of the generative AI world (July,5th 2023)

Details: Written by: Super User; Category: news; Published: July 05, 2023; Hits: 1515

- Exllama an implementation of Llama has been released. According to the author, it is "designed to be fast and memory-efficient on modern GPUs". And it seems to be able to handle longer contexts. (A test will come)

- AudioPaLM is a project from google that allows to convert speech to another speech in a different language and even to change the accent of the speaker. The link links to a example.

- DragGAN has finally been released. As explained before, DragGAN can modify a picture thank to some kind of vectors of transition (it is me who call this like that)

- Zeroscope V2 is a text to video generator. Click on the link to test the demo on Huggingface. Click here to get a list of the demos of the same user in Huggingface. (There is a way to add the parent projects is Automatic1111).

Stability AI announces SDXL 0.9

Details: Written by: Super User; Category: news; Published: June 28, 2023; Hits: 1527

Stability AI, the company behind Stable Diffusion, announced the release of SDXL 0.9, a new diffusion software (like SD 1.5 or SD 2.1), in mid-July.

If you click on the thumbnail above, you will be redirected to the video announcement of the Youtube channel "Stable Foundation" about the release of SDXL 0.9. The 49 minutes video explains the roadmap of the development of the software and the related models. The specification that one needs to have on his computer to actually run it and more...

If you want to read the official announcement, on the website of Stability AI, click here.

If finally, you want to test SDXL 0.9, you have 2 options:

You can go to ClipDrop for SDXL 0.9 (free, without registration but with a queue).

You can also go to Dreamstudio for SDXL 0.9 (free, requires registration, limited number of credits).

(and Btw the video that i link above says that these 2 options will likely be merged in the future)

Easy deep fakes with Roop

Details: Written by: Super User; Category: deep fake; Published: June 27, 2023; Hits: 2953

Make deep fake pictures and videos with Roop

Roop is a project to create deep fake contents.
It exists on Github as 2 different projects:
- One is a standalone project to create deep fake videos, but also pictures.

- The other one is an extension for stable diffusion's interface Automatic1111 and it allows to create deep fake of pictures.

Some links to generate music using an AI - text to music

Details: Written by: Super User; Category: Music Generation; Published: June 10, 2023; Hits: 2856

If you are familiar with text-2-img picture generation, here are some links to generate music, locally on your computer with an AI, using some lines of text:

https://github.com/facebookresearch/audiocraft

Audiocraft by meta/facebook research

https://aitestkitchen.withgoogle.com/experiments/music-lm

MusicLM with the AI Test Kitchen demo

Moûsai:

https://github.com/archinetai/audio-diffusion-pytorch

Riffusion:

https://github.com/riffusion/riffusion

Tests and installation comments are on the way.

H2OGPT: Another tool to ask questions about your own documents but on GPU!

Details: Written by: Super User; Category: Text generation; Published: May 29, 2023; Hits: 2013

H2OGPT: Another tool to ask questions about your own documents

H2OGPT is related to the project H2OAI is a Web UI (which means web user interface) a bit like the oobagooba web ui.
In a matter of a saturday, this project went in my own ranking from "barely usable with poor result" to "pretty decent".

Clone your voice on a song

Details: Written by: Super User; Category: Cool stuffs; Published: May 25, 2023; Hits: 1232

text to speech

If you are looking for a tool to use the voice of someone else on a song (a bit like that), click on the thumbnail above to watch the tutorial and give it a try. I have not tested the tool myself, yet.

Tutorial: How to Interrogate an AI about the content of a youtube video locally?

Details: Written by: Super User; Category: Random ideas; Published: May 25, 2023; Hits: 1785

This tutorial is going to use PrivateGPT as a tool to interrogate a youtube video. Since PrivateGPT is able to read documents, if we can convert a youtube video to a document, we can interrogate the youtube video.

You have been rickrolled - AI version

Details: Written by: Super User; Category: Cool stuffs; Published: May 24, 2023; Hits: 1270

While it is still not very easy to create videos using AI (like text-to-video), the result of some people is quite interesting.

If you click on the thumbnail above, you will be redirected to an AI version of the clip of "never gonna give you up".

Popular articles

Home