Description
Image imported from Derpibooru.
Original: https://derpibooru.org/images/3008441
Metadata at the time of import:
Original: https://derpibooru.org/images/3008441
Metadata at the time of import:
| Field | Data |
|---|---|
| Favorites | 129 |
| Upvotes | 207 |
| Downvotes | 7 |
| Score | 200 |
| Comments | 10 |
| Uploader | EpsilonWolf |
Original description below this line
More comfy images done by V3 Purplesmart.ai. Check out the discord for more info
Prompt:
Princess Twilight, alicorn pony, cute smiling pony face, detailed mane and fur, chest fluff, happy smiling, perfect anatomy, extended full wingspan, standing in grass meadow, beautiful, soft shadows, cinematic lighting, smooth gradiants, masterpiece, professional digital art by artgerm and Greg rutkowski, p_95, cutie mark:1.3
Negative prompt:
noise, film grain, ((deformed)), blurry, ugly, ((mutated)), extra limbs, ((poorly drawn face)), poorly drawn, low quality, ((out of focus)), signature, username, p_low
Settings:
Steps: 35, Sampler: Euler a, CFG scale: 8, Seed: 19429457, Size: 1024x1024, Denoising strength: 0.5
what does bother me about AI drawn ponies quite often tho is the lack of any manual post-processing/corrections or detail work by the uploaders. Too many of them it seems would rather spend another hour looking at new attempts by the AI instead of taking a good image into Photoshop to correct the last few errors themself. It’d be quicker and even a good way to train the AI on the final image, but why bother doing anything at all.
On the other hand, and as unfair it may be toward some of the great traditional/digital artists, images that did receive a lot of post work and are more “machine learning assisted” than entirely ai generated are among the best pieces on this site. Well, art is subjective anyway and as I like to say: There will always be as many sides to a story as there are people telling it … ^^
AI “art” takes no skill or style to make, anyone can become an artist if AI art is considered as real art. Even someone who has never drew anything before. Art isn’t just looks, skills and creativity is a major part of it as well and AI lacks both of those elements. The real purpose of AI “art” is to generate quick images that are needed for either commercial or advertisement purposes or any other purpose that requires images in a quicker pace than a human can draw.
Art has several different components, the foremost is the technique and execution to make an aesthetically pleasing image. This component is the easily recognizable element (and most sought after) in art and thus is often reduced to this component. You are correct that AI has been able to grasp the vague concept of what a good image should look like and is able to create it very well. The other components are the use of complex, abstract concepts to enrich the meaning of the piece and the message (or emotion) that the pieces inspires in the viewer. These latter two are (currently) missing from AI and is what makes human-made art “unique” and different.
However, most of the art uploaded on DB relies solely on the primary element of art: how it looks. From a pragmatic perspective, AI images can be considered art since 1) it is able to realize at least one of the key attributes of art and 2) it’s art fits the qualifications that most artists have already chosen to pursue: making pretty pictures.
I think this speaks to how both AI and human artists are also severely lacking in the more complex (and impactful) features of art; AI can not do such features, and human artists have chosen out of their own volition.
There is a general misunderstanding in how (AI) Neural Networks and stable diffusion work. As a broad blanket statement, Neural networks function and learn in much the same way people learn; Through repetition and a diverse range of examples, both AI and people learn to associate certain qualities, concepts, and the literal process to accomplish a set goal. In much the same way a person must go to a class to be shown the what, where, and how to draw [something] in a particular way, AI must learn through examples and exercises. A person who has never heard or seen what Impressionism looks like, can not draw art in that style. Furthermore, practice and examples are needed before a person can begin to replicate that particular style.
More intermediate thinking like composition, architecture, “world building”, environment, etc. are concepts that AI does not do well. At the very least, AI can be used as a tool that solves time consuming and boring work that detracts from what the artist wants to draw, such as backgrounds, building, perhaps even crowds or basic scenery items. Human creativity is unrivaled, and we should push ourselves to bring about fresh ideas and new concepts that are more impactful and lasting.
Because the majority of people would rather chose to support a machine that rips off real artists work rather than a human being who works for years on art to achieve a good art quality level.