Rhodynoliaeth

I want to update my biography with more current and general content. I will link my old bio here. For people who are more curious.

Hello, my name is Gael, but you can call me by my username Rhodd. What motivates me to publish and interact is to meet people with whom I could potentially build connections. I don't care if you're on the other side of the world; I believe in humanity, even though I know that on such a vast platform as the internet, there's a higher chance of differences. That's why I believe art provides a window into the inner self of others, and that's why I give Deviantart so much credit.

I'm very introverted in real life. I feel more comfortable in a place where I can think and put my words down rather than direct social interactions. This explains why I'm quite alone in my everyday life and a bit rough around the edges. I consider myself progressive in the Marxist sense, trying not to harm minorities and not encourage sexism, xenophobia, or mistreatment of LGBTQI+ individuals.

I remain a cis white man born in a postcolonial country. Paradoxically, I love themes related to sexuality. I don't like hiding it because it's a lot of relational work to make this aspect understood. But I don't like the humiliation aspect. I admit that publicly asserting these things may alienate some people, but I think it's more beneficial if people know upfront what they're getting into by being honest from the start.

But then you might say it's not evident at first glance. No, my gallery has taken a nature-oriented direction as the main theme, not by editorial choice but because of my ability to create art.

My early experiences were in Photoshop as a photo manipulator, creating surreal scenes with a thousand colors. I started shortly after turning 20, never having been involved in art. This might explain it; I'm not an art genius, just a modest and average artist.

Then I discovered photography, and that's when I gave myself to nature. I grew up in an industrial city in the petrochemical industry but spent my vacations with relatives in the mountains of southern France, which influenced my childhood. That's where my inspiration primarily comes from.

Today, I've moved to these mountains, and for someone not very social, it didn't change much for me. Except that I now live in a dream place for me, far from the city's conveniences but close to dreams of waterfalls, small wild animals, leaves of all colors in autumn, and a thick white blanket of snow in winter. And the meadows with a thousand colors in spring.

Then I pursued training in graphic design, in the general sense, producing communication materials to meet the needs of businesses to promote their services or products. I know a lot in art theory, how to do things right in composition, hierarchy in drawing, the sense of reading. From those years, I learned to draw, but I'm not really talented. It's a great frustration.

I would have liked to create stories, draw contexts and ideals, express my carnal desires, my love for the female body, and my plea to respect the desires of others and their mental and physical identities.

But that's life, and that's where artificial intelligences come into my life. I don't talk about it much here because this place can be elitist, and the guardians are often self-centered. This topic is very divisive. So, I don't take on the identity of someone trying to create imaginatively under AI. But it remains my underlying goal. I try to humanize the machine by doing what most traditional artists don't do well – being imaginative. This extends to the thought process, creation, and execution.

I use my years of design studies for this. The most important thing I was taught is not to underestimate idea proposals. Don't rush in headfirst when you have an interesting idea.

This is what sets me apart from creators who churn out tons of Waifus. I put all my knowledge into creation, and I try to make you dream with it, all while expressing my identity and satisfying my desires and sharing them with you.

So, now you know the big picture. Thanks for reading. Be yourself, respect others, and most importantly, take care of yourself!

I'm pondering a question... Should we welcome artists who don't use AI? Honestly, with all the hate directed towards us and our medium, should we make the same foolish mistake as those who lack any soulful openness? No...

I believe I will evolve... And propose a portfolio and accept friends of Synthographers that we are. This will help gain visibility for our group. And work towards unifying ourselves with respect and for the love of creating and sharing our ideas...

However, the foundation that identifies Imagine AIart will remain the same for them; we are all limited in the number of daily publications because I want to share public visibility with as many people as possible.

And I want to hassle you to train yourselves to sort and make choices about your works. To fuel your imagination... We don't all solely rely on AI, and it's also good to showcase your other talents...

So, I invite you to this new practice... It does mean, though, that creation via AI still remains our main theme and I will, as usual, be strict with portfolios for those who don't create with AI. The folder : https://www.deviantart.com/imagineaiart/gallery/91726394/synthographers-friends

Here is the message you may receive when you apply for admission.

Hello, recently we changed our editorial line, we now welcome artists from all backgrounds under the following conditions:

This group remains focused on illustrations generated via AIs, but you are entitled to a folder to participate in your own way in sharing the love of art, showing imagination to modestly change the world on our scale. Here is the folder : https://www.deviantart.com/imagineaiart/gallery/91726394/synthographers-friends However, you are subject to the same rules as the Synthographers in the group, meaning two publications per day.

And you must find a different theme for each publication, and unfortunately, I will reject your submissions if they are not in the right portfolio. One last point, if I see you disparaging the Synthographers anywhere on the site, it will be normal for me to show you the door... If all of this suits you, then you are welcome?

So, if you are curious to set yourself the challenge of renewing yourself in the themes you address, then this group may be of interest to you.

For my part, I benefit by expanding our community with you friends of the Synthographers. So don’t hesitate to offer your help and your open mind! ImagineAiart

#ai #aiart #aiartcommunity #synthography #synthographer #illustration

Hello folks!

I've completed the first phase of my creation. It's often the trickiest part since I'm limited by the training the models I use receive. Even though model creators do their best to make them flexible, thinking outside the box is where it gets really challenging, sweating for days just for a damn sentence...

As I explained in a previous post, contrary to popular belief, it's not enough to type a sentence while drooling and half-asleep with your hand in your pants...

I took half a day to think about my illustration, digging around the internet and searching for ideas here and there.

But for three days now, I've been struggling to generate a decent image that simply reflects my concept. Why? Because you guys think linearly, and in practice, it doesn't work that way. If I just wanted to show you a naked woman, it would be simple because the models are primarily trained to generate portraits. Now, in my case, I'm doing staging. So, I'm asking a model to use its knowledge to create a composition it has never been trained for.

I'm asking it to create a landscape by adding elements it usually only uses in a secondary role, typically not as the main subject. And that... it doesn't like very much. To convey the importance of an element that we want to highlight in the illustration, you have to give it an order. For that, you have arguments to support the words.

"A cat and a ball of yarn."

With arguments: "A cat++ and a ball_of_yarn--"

Another variation: "A ((cat)) and a [[ball of yarn]]"

Another variation: "(A cat:1.3) and a [ball of yarn:1.3]"

In these three examples, I asked the AI to represent the cat more than the ball of yarn. You can imagine a low-angle shot with the cat huge in the background and a tiny ball of yarn in the foreground.

But these are easy elements to represent because models are often placed in situations with animals.

Now imagine that you are without limits of imagination and you would like to see the Eiffel Tower wrestling with the Statue of Liberty. These are two very specific concepts observed only in the context of passive and inactive elements. Wanting to make them adopt another concept is against nature for the model, and that's where we go beyond the simple fact that AI does all the work for you. If people were content to just follow the rules, we would only have subjects posing and being passive.

For this, we have sub-programs that are executed within the main program. For example, Controlnet allows, starting from a skeleton, to give poses to our characters to help the AI imagine the context of the positions it should represent in our illustration. This can help us put the Statue of Liberty in a combat posture. We could also use a variant of its features by drawing a draft of the Statue of Liberty just to show it the position it should represent.

By the way, consider this as a concept just to have to explain this to the AI. Why dwell on the concept principle? Because the AI is not capable of dwelling on multiple ultra-detailed concepts. The more time you spend detailing and explaining how you see things, the more it will be unable to detail other concepts. Overloading your AI results in creation bugs, like blur, abstract shapes, and an ugly illustration. That's why it's easier to make it understand suitcase concepts like a cat playing with a ball of yarn. Unless your model has been specifically trained with images of the Eiffel Tower and the Statue of Liberty fighting.

That's why typing a simple damn sentence can take three days... because I wanted three very specific concepts, and they are incompatible. And because I like to bring my imagination to life. And in my case, it's a landscape with an object that you don't find in this context, and it was difficult not to contradict the AI too much so that it wouldn't panic and produce a crappy rendering. And generating an image is not instantaneous, it's long and boring, it can take 2 minutes per generation, and it's just tests to see how well the AI understood the given orders. Here's a 6-image example of a single concept and how much you have to "yell" at the AI to get it to listen to you on a concept that's basically foreign to it, so imagine with 3 concepts at the same time.

The more you "shout" orders at the AI, the more clumsy it becomes... There you have it, I hope you understand our work better, and I hope that by sharing knowledge and not insults, you understand that we demand our place on an art site too. Even if it's a universe slipping away from you... And yeah... AI are not artists, but we are, whether you acknowledge it or not. Just like classical music elitists do not recognize and discredit rap, we are artists with a message to convey. Even if it's just the right to express ourselves and claim the title of artist, even if it irks you. We want the same chance as you to benefit from the visibility that the site's algorithm provides. If it bothers you, then create your own tag for 'true artists' and leave us alone! #ai #aiart #aiartcommunity #aiartists

Hello, I want to share with you my way of working on a digital illustration. But also, my habits and hobbies through this medium.

Since Aug 7, 2022, things have evolved a lot. I migrated from so many mobile applications to an optimized workstation just for this. I never thought I would have such expensive equipment mainly for this.

We're talking about a Microsoft Surface 8 Pro 16GB and an i7. The tablet went from a 512GB SSD to 1TB, and over time, I invested in 3 SSDs of 2TB each, totaling 7TB of storage space. It fills up quickly with large models. It's a very large file that contains hours of training data and instructions to make sense for the software that generates images from these models.

A model cannot be precise in the shapes it generates with the orders it receives and be very precise in its ability to follow orders and handle multiple ultra-wide domains, such as portrait, landscape, photorealism, manga style, ComicBook style, macro photography, and more.

A model has a file extension named Safetensors or Checkpoint, which is a folder filled with reference images, each with a large table used to give references to the image generator that will use them as a reference. Creating a model takes hours, and it takes months to refine a model to be the most accurate and responsive to the orders given in the image generator.

It's truly an art to create the most popular model in the artificial image creator communities, and the most popular ones become stars in the community.

A model can weigh between 1GB50 and 12GB, so a hard drive can quickly fill up. It's my favorite hobby to look for a completely unknown model that gives a crafty and original style that I love. I can spend days just organizing all the models I have. A waste of time? No, it's like searching for a stock of images on the internet, cleaning up, and preparing models in advance for Photoshop to do photo editing. We don't settle for cutting out each time; we need to organize to save time every time we reuse a cut-out image.

It's roughly the same approach with my work. I make categories of models that perform well in a specific domain I tested them in, like portrait or landscape, realistic or anime, nude or safe. This way, when I start a new project, I know what style to choose that will better fit the expression I want to give to an illustration.

And just that, it's research and testing over several days, which is what I prefer. Discovering little-known worlds is a bit like opening the wardrobe in Narnia movies, but each world is different to the eye.

It's an idea I just had. It's simple, Google street is the answer. With img2img and a good prompt you can move wherever you want.

Mmh, small self-criticism.. I think that the images used belong to Google. So well... Ethical level...