Reviewing Leonardo.ai

App.leonardo.ai is an innovative Image content generation process assisted by a powerful artificial intelligence (AI) offering a wide range of features and capabilities.

The interface is modern, very much inspired by discord in the Web 3.0 style

In terms of performance, app.leonardo.ai delivers impressive results. For people not able to run SD at home, Leonardo is a good way to access several good quality models such as SDXL, RPG4, RealisticVision, Dreamshaper, etc ...

To access Leonardo.Ai, you can simply sign up by visiting their landing page and clicking on the ‘Create an account‘ or ‘Start using Leonardo‘ button. 

If this is your first time signing up, enter your first name and email and click the ‘Count me in’ button to be added to the early access waiting list. 

However, if you’ve already been granted access, you can click the ‘Yes, I’m whitelisted’ button which will take you to the web-app login.

Generating images will cost tokens (and the daily amount is 150) by tweaking the settings generating can cost as low as 1 token (disabling Prompt Magic, and Alchemy) otherwise a generation can cost 16 token, on a free plan all generated images will be made public.

An access to the community pages will list all generated images and if one catch your eyes it can be used as a base for a remix, or img2img ; other options such as 'remove background' or 'upscale' are also available.

Here is an example of a generated image:

with this settings:

Full body Portrait of Beautiful toned Megan Fox, high quality illustration in post apocalyptic , muscular, messy hair, award winning photo shoot 

and model 'Dreamshaper v7'

Meet John Doe (1941)

Thanks Ron Perlman for the recommendation 

Converting photo into anime shot

Stable diffusion can be used as generative engine starting from a prompt, but it can also be used as a filter of sort, adding a photo of Jenna Lynn Meowri as input and setting a filter as a Lora can give pretty fun results 

Photo credits Linktr.ee/jennalynnmeowri

The state of the AI vs classic artwork discussion

The introduction of AI technology into the artistic sphere garnered significant attention a few months ago, inciting heated debates, lawsuits, and controversies. However, the novelty of this trend appears to have diminished recently.

I can see some artists still balancing from the old ways and now including AI into their workflow

One artist in particular caught my attention, he seems to be swinging back and forth, posting drawings and then AI artwork 

Here is the link to his blog blogclif.wordpress.com/

Here is a recent sample of his drawing

and here is the AI generated from the drawing

My current view is that he should create numerous drawings and select the best ones to evaluate their appearance in the AI artwork style.

Deviantart and AI Artworks

DeviantArt is a popular online platform that allows artists and creators to share their work with a global audience. 

The website has a vast collection of digital art, traditional art, photography, and other creative works, making it an excellent resource for inspiration and networking. 

DeviantArt also has a strong community aspect, with groups and forums dedicated to various art styles, mediums, and interests. This allows users to connect with like-minded artists, get feedback on their work, and collaborate on projects. 

Additionally, DeviantArt offers various tools and resources for artists, such as tutorials, stock images, and brushes, making it a valuable resource for both beginners and professionals. 

Recently with the AI trends DeviantArt has implemented an innovative approach to AI-generated artwork. The website has a feature called "AI Generated Art" that allows users to upload artwork that has been created with the help of AI algorithms. 

This includes artwork generated with neural networks, machine learning, and other AI techniques. The artwork is tagged as "AI Generated" and is given a unique copyright status, which allows artists to retain ownership of their work while acknowledging the contribution of the AI algorithms. This approach is designed to recognize the role of AI in the creative process while ensuring that artists are properly recognized and compensated for their work. It's an interesting and forward-thinking approach that highlights the potential of AI in the creative arts.

Overall, DeviantArt is a fantastic platform for artists and anyone interested in the creative arts, and is definitely worth checking out.

But there is this little trick I found quite useful, when wanting to credit an artist for a picture, after downloading an image, you will in the file name a code that can be pasted into an url shortener and get back to the place it was taken from.

for example this file 'just_a_cute_cat_in_a_red_bag_by_trumanbrown_dfjtqpo-pre'

correspond to this picture :

Like this picture

it contains this code 'dfjtqpo'

and when adding this code to the url shortener fav.me/

like so fav.me/dfjtqpo

you will be redirected to the creator's page 

Pretty neat

Taking AI animation to the NEXT level

Comparison portrait combining Scarlett Johansson, Kristen Bell, Christina Aguilera, Angelina Jolie, Heather Graham

 A while ago this question was asked on this blog

What if a crazy scientist would create the perfect model by combining :

Scarlett Johansson

Kristen Bell

Christina Aguilera

Angelina Jolie

Heather Graham

Now with the assistance of AI (stable diffusion) and this website

it is possible to get several portraits:

The prompt used:

one perfect female face mixed of (Scarlett Johansson, Kristen Bell, Christina Aguilera, Angelina Jolie, Heather Graham), looking at viewer 

Downloading songs as mp3

Fetching mp3 from youtube video on Ubuntu Linux

the selected tool is 'youtube-dl'

from the terminal prompt write:

$pip install youtube-dl

but then you get this error (because we are March 2023 and things are broken)

ERROR: Unable to extract uploader id; please report this issue on https://yt-dl.org/bug . Make sure you are using the latest version; see  https://yt-dl.org/update  on how to update. Be sure to call youtube-dl with the --verbose flag and include its complete output.

for the long version of this error:

ERROR: Unable to extract uploader id; please report this issue on https://yt-dl.org/bug . Make sure you are using the latest version; see  https://yt-dl.org/update  on how to update. Be sure to call youtube-dl with the --verbose flag and include its complete output.

    'uploader_id': self._search_regex(r'/(?:channel|user)/([^/?&#]+)', owner_profile_url, 'uploader id') if owner_profile_url else None,

  File "/home/_dl/extractor/common.py", line 1012, in _search_regex

    raise RegexNotFoundError('Unable to extract %s' % _name)

youtube_dl.utils.RegexNotFoundError: Unable to extract uploader id; please report this issue on https://yt-dl.org/bug . Make sure you are using the latest version; see  https://yt-dl.org/update  on how to update. Be sure to call youtube-dl with the --verbose flag and include its complete output.

To fix this error you will have to type this:

$ pip uninstall youtube-dl

then type this:

$pip install git+https://github.com/ytdl-org/youtube-dl.git@master#egg=youtube_dl

but that means you need to install git

$ apt install git 

will do

while you're installing new package don't forget to get ffmpeg as well

$ apt install ffmpeg

Now you're ready to get the song you want

Art and Sponge bob

How to download a ckpt model

These ckpt files are huge (at least 2GB and usually 4GB, they can reach up to  8GB) 

First method:

1. An easy way to get them inside of a google colab

by adding these commands inside a code snipet:

!git lfs install
!git clone https://huggingface.co/WarriorMama777/OrangeMixs

git lfs install is required to download file heavier than 100MB

and the clone command is a way to copy the huggingface.co repository 

in this example the OrangeMixs model

2. Alternative way

Using the huggingface-hub library for Python

!pip install huggingface-hub
from huggingface_hub import hf_hub_download
hf_hub_download(repo_id="WarriorMama777/OrangeMixs", filename="VAEs/orangemix.vae.pt", cache_dir="./")

Preparing a LoRa (low Rank) dataset

In my previous article, I asked to prepare a dataset of a few pictures, to train a LoRA model

This is actually paramount for successful results later, and therefor I will go deeper into that process

First gather about a dozen pictures of the same subject, this could be a character, a celebrity, or yourself (to create your avatar)

choose various background, various clothes, various face expression, angles etc ... the more divert the better

then resize the pictures to 512x512 pixels, and save it with no too much quality loss 

the name of the pictures should be 1.jpg, 2.jpg, 3.jpg etc ...

After this is done, open a text editor and create 1.txt, 2.txt, 3.txt for each pictures

In that file describes the pictures, using a keyword that will be the LoRa keyword (its name) make it simple and unique, something like Gdlf for Gandalf or Hptr for Harry Potter 

then add the variations made in the picture, the hair style, the background, the angle, the clothes etc ... separated by comma.

Here is an example:

1.jpg is

1.txt will contains:

cbpk, white hair, looking at viewer, leather jacket, yellow abstract background

The LoRa will be cbpk.safetensor (for Cyberpunk)

Training a lora (stable diffusion) [part 1]

Step by step Lora Training

Training a model is very demanding, and the result is a minimum 2GB file ; while a Lora (Low-Rank Adaptation) will be less than 150MB.

Where to start?

I would recommend this google colab link

Linaqruf/kohya-trainer/blob/main/kohya-LoRA-dreambooth.ipynb 

Step 1:

Requirements: The preparation of a set of images, and texts (dataset)

Another requirements is a google account (to use the colab link and google drive)

Each image and text's name must match - the size of the images should be 512x512 pixels 

It is convenient to use https://www.birme.net/ to resize the dataset.

For the text, best is to give a little description on each file, or to use the img2txt feature and get the terms directly from the AI.

 Only 5 images (and texts) are going to suffice, but 15 or more will make the model more reliable.

Once the dataset is complete, put these files into a folder with the number of repeats (5 or 10) and the concept (something like exp, or person, or Bcloth, etc ...)

Once ready add the datasets into the google drive

Step 2: 

Open the colab trainer

Go to github.com/linaqruf/kohya-trainer

and pick Kohya Lora Dreambooth

To be continued ...

Download CIVITAI Models inside Stable Diffusion webui

Energy takers vs givers

• Spending time outdoors
• Eating good food (organic)
• Listening to classical Music
• Living mindfully
• Proper diet (no junk food)
• Creating stuffs
• Or cleaning

Creating content is so easy now

Cartoonists, musicians, voice actors, and more broadly every artists or content creators are now in competition with AI generated contents.

This is a game changer, and we have not completely understood or contemplated how much of a revolution we are seeing emerging ... in early 2023

It took me less than 10 minutes to create a new super hero from scratch, write a backstory, and several scenarios, in less than a day, I could self publish an entire comic book featuring this new Dark violet super hero ...

I used lexica.art, dreamlike and stable diffusion to get a different artstyle

lexica.art is crazy good

exica.art/prompt

prompt:

highly detailed portrait of a sewer punk lady student, blue eyes, tartan hoodie, white hair by gustav klimt , art neuveau, atey ghailan, by greg rutkowski, by greg tocchini, by james gilleard, by joe fenton, by kaethe butcher, gradient yellow, black, brown and pink color scheme, grunge graffiti tag wall background

https://lexica.art/

Dreambooth Tutorial SD1.4

Run Stable Diffusion with your mac in one click

Everything is easier on a mac

once again it is...

download at https://diffusionbee.com/

DiffusionBee needs:

• Apple M1/M2 
• 16GB or RAM
• macOS 12.5.1 on higher. 
• Machines with Apple silicon are recommended.

Transform Your Selfie into a Stunning AI Avatar with Stable Diffusion - ...

Tsu interview : Anna Lioce (aka MisaLioce)

Q1: Can you introduce yourself (give a short bio) ?

I live in South Italy and I am a geek girl.
I love drawing comics with my husband (he is my favorite writer) and watching SCI-FI and horror series (always with him).
I've been drawing since I was a little child and I hope one day to become a professional comic artist.

Q2: What convinced you to join tsu.co?

I joined tsu because I'm interested in new media (I am a graphic designer and Marketing expert too).

Q3: What are the main tools of your trade?

I use Manga Studio and Photoshop.

Q4: What are the biggest struggles you encounter as an artist?

People!! People thinking that drawing is something easy and/or stupid!

Q5: What are your visions of art for the future?

For me, art will be web-based and shared with the world and other artists, giving people the unique opportunity to know other cultures and way of expression

Q6: Where can we find you on the web?

twitter : https://twitter.com/Misa0027
tsu: http://www.tsu.co/MisaLioce
Ghostbusted comics : http://www.staynerd.com/nerdstuff/stay-nerd-comix-attenti-a-quaynerd-06-ghostbusted-finale/

Thanks for the interview Anna Lioce,

you can join her network at http://www.tsu.co/MisaLioce