Updates in the image generation tier and price increase. (Patreon)

First I would like to advise that no current subscribers will be affected by the new prices if they do not renew their subscription.

Allow me to explain why.

Many of you may know that the tier was initially launched with a limit of 20 spots, then reduced to 5, and later increased to 10.

Perhaps many of you might think, especially with apps of this kind, that generating images is done with a click of a button, but this service is not about that.

Firstly, the point is that I offer 4 themes (ideas) to each subscriber. Unlike common apps, I don't offer the first thing generated; I refine the generation through many iterations. When an image reaches a minimum standard of quality, an intensive work process begins on that image or images. This involves reconstructions, inpainting, retouching in Photoshop, and other adjustments to achieve extremely high-quality images with very high resolution (up to 4K wallpapers).

The reason for the price increase is simply that it takes me more than an hour for each theme (out of the 4 per user), depending on how much manual work is required to refine an image.

The image accompanying the post is a generated image for a subscriber of the tier (thank you very much), and I will use it to illustrate the issues I'm talking about.

For example:

What you are seeing is how the original background characters were generated on the left side, and how they turned out after manual fine-tuning.

I'm sure many of you also know that AI struggles with hands, so they pose another challenge that needs to be addressed with a combination of fine-tuning and Photoshop. (deformed hands, missing fingers, etc. In this case, the hand was slightly deformed and had only 4 fingers, furthermore, it didn't even hold the feather correctly; it actually passed behind.)

Similar case for the other hand.

By now, I believe you all understand perfectly that I put care and love into each image to ensure it has at least an acceptable level of quality.

Then there's the challenge of generating them in very high resolution (with minimal or no upscaling). In this case, the final image is 2560x1080 (slightly higher in reality) without any kind of upscaling.

The tier now has a price of €15 for 3 complete topics.

However, that doesn't mean you will receive only 3 images.

In this case, this image corresponds to one theme, but I offered 17 images in total. There were 11 interesting intermediate-sized generations and 6 full-sized ones (where 3 correspond to variations of the main idea). Variations in face and expression are perfectly valid, you can request them at no additional cost.

If you know a bit about image generation, you will also be interested to know that I use several different models depending on the desired style. I utilize techniques such as inpainting, high-resolution fixing, Lyco, Lora, and various scripts in a specific workflow. This workflow consists of predefined steps and involves the use of certain Lora functions, such as "add_detail," to subtract or add detail to sections that may unbalance an image.

Therefore, the images I generate, after undergoing steps in img2img and Photoshop within my workflow, do not correspond to any specific prompt. They cannot be reproduced automatically (which is why they require so much work).