Tumgik
#ai portrait
slick-devon · 11 months
Text
Tumblr media
Light and shadow
88 notes · View notes
gingerbredman1989 · 2 months
Text
Tumblr media Tumblr media Tumblr media
Portrait of a working-class man, circa 1895.
Bing AI Image Create
19 notes · View notes
corre1310-blog · 5 months
Text
Tumblr media Tumblr media Tumblr media
Toying around AI Art using Bing;
Create a "Muslim Mortal Kombat character" and apparently AI ONLY THING DETECTED MK SCORPION AS MORTAL KOMBAT CHARACTER PROMPT
BTW,Idk should I consider it as one of my Mortal Kombat original characters; Aim as she's A Japanese Muslim Shirai Ryu that has samurai ancestor that using firearms instead of sword
BTW, it's beautiful
23 notes · View notes
goodeyeforawhiteguy · 11 months
Photo
Tumblr media
26 notes · View notes
hombresguapos · 14 days
Text
Tumblr media
Retrato de un hombre treintañero Imagen generada por IA
3 notes · View notes
totaldramabyrinth · 6 months
Text
Tumblr media
When AI makes you want to change your whole aesthetic.
7 notes · View notes
disordinarybeauty · 1 year
Photo
Tumblr media
DISORDINARY IMAGE-TO-IMAGE
Did image-to-image outperform text-to-image in my quest to find out what DISØRDINARY BƏAUTY 🥀 ⚡ BÆUTY IZ CH∆ØZ ⚡ would look like if it was made by AI instead?
The collection project in discussion is DISØRDINARY BƏAUTY 🥀 ⚡ BÆUTY IZ CH∆ØZ ⚡, an experimental exploration of atypical beauty found in the subject of depiction studies focusing on the weird, the abnormal, the creepy, the imperfect, and the ugly ones. As an artist, I am currently investigating how AI is influencing our sense and aesthetics of beauty in art and the perception of beauty and ugliness in society. In pursuit of this, I am conducting experiments with AI and the use of prompts for text-to-image to discover what the collection DISØRDINARY BƏAUTY 🥀 ⚡ BÆUTY IZ CH∆ØZ ⚡  would look like if it was made by AI instead.
In this context, the image-to-image approach appears to be a more efficient way to investigate how the Disordinary Beauty collection would look if it were made by AI. While the text-to-image method offers a broader range of possibilities by generating images based on a prompt such as "Generate a portrait in the style of Disordinary Beauty by Italian glitch artist Domenico Barra" the resulting works are the products of AI "imagination" and "interpretation."
However, the computer vision approach, which uses an original image as a reference, produces results that are more faithful to my original works but less "authentic" in the context I am exploring. To answer the question of what Disordinary Beauty would look like if it were made by AI, the text-to-image approach is still my preferred method, as it better reflects the AI's creative process and interpretation.
The use of AI in the art world poses complex questions about authorship, authenticity, and creativity. The DISØRDINARY BƏAUTY 🥀 project aims to examine these issues and explore the intersection between AI and human creativity. As such, the findings of this experiment invite us to elaborate on the significant implications for the future of art and its relationship with technology and how this could impact the work of artists.
Tumblr media
These first two sets of images were generated in an AI image system app for mobile devices, I used Android, and the app is called Starry. AI. This app runs on Stable Diffusion. This is my favorite app so far because I can use it on my mobile everywhere I am and whenever I need it. Anyway, in these two sets, I have used as a prompt the artwork shared at the top of this article.  
For this first test round, I used the artwork. ㄩ. 几. Ҝ . 几. ㄖ. 山. 几 .22 (the image shared at the top of this content page).
In the first set, I opted for the “Create Art” model, in the second one I opted for the “Create Photos” one. In both options, the user is required to also include a text prompt which will be the actual answer to the command “What do you want to see?”. The text I used is the root copy I am using in this investigation; “A glitch art portrait in the style of Disordinary Beauty by Italian glitch artist Domenico Barra”. The image uploaded will only work as a guideline for the style, so the structure, shape, colors, etc. Tendentially, Starry.AI in “Photos” mode will almost always deliver a subject with an attractive face, which conflicts with the  DISØRDINARY BƏAUTY 🥀 aim to artistically redeem the “ugly”. The art mode manages to deliver images that I find more inherent to the core of the project.
I like that Starry.AI gave facial features to the images, so now I could pick a look for my original faceless “freak”! :)  
Tumblr media
In this set of four, I used DALL-E 2. This AI system is my least favorite one when it comes to text-to-image because it usually delivers very cheesy glitch art-like images but I like how it handled these “promptless” image generations. DALL-E 2 does not require a text input, it works directly based on the image sample provided. These are very hauntingly beautiful portraits, and this fit in the collection project claim, “the thorn sense of beauty”. 
Tumblr media
This set is generated by Runway ML. This AI system platform is very powerful. In image-to-image requires uploading an image as a reference and also a text prompt. The results are very far away from the original image, with way too many colors, and I do not see much difference from when running a text-to-image model using the original AI prompt for DISØRDINARY BƏAUTY 🥀. This system also gave a “face” to the artwork. 
Here I share the same process and the same AI Systems tested on another artwork part of DISØRDINARY BƏAUTY 🥀 ⚡ BÆUTY IZ CH∆ØZ ⚡.
Tumblr media
 This is the original artwork: . ㄩ . 几 . Ҝ . 几 . ㄖ . 山 . 几 .17
Tumblr media
In this set, Starry.AI went completely freestyle and generated four images not coherent with the sample provided. The reason I believe this is that Starry.AI always associates the word “beauty” with female figures, so this is why in this set it delivered four female portraits. Also, three out of four characters stare towards the right instead of the left, holding a right position. 
Tumblr media
DALL-E 2 did a good job on this second test, probably the one delivering the best solutions compared to the other two. Also here the characters hold the right position, different from the original artwork. The images are still pretty haunting, I like them. 
Tumblr media
Runway ML proved once again to be the wildest one! :D
12 notes · View notes
originalsdesigns · 2 months
Text
Tumblr media Tumblr media Tumblr media Tumblr media
Retrato de mujer con flores a estilo fantasia
--------------------------------------------------
Portrait of woman with flowers in fantasy style
* Comment what you think
* Value my work giving a like and follow me for more content
6 notes · View notes
peach-point-avenue · 1 year
Text
Tumblr media
13 notes · View notes
aidreams23 · 7 months
Text
Tumblr media
52 notes · View notes
slick-devon · 11 months
Text
Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media
3,000+ Follower Mega-post! Thanks for the encouragement over the last few months! Finally, I believe I'm getting details, consistency, and realism. To celebrate, I upscaled these overnight. Also, happy Pride Month!
67 notes · View notes
gingerbredman1989 · 2 months
Text
Tumblr media Tumblr media
Portrait, oil on canvas, the subject is a 21-year-old male bodybuilder.
NightCafe AI
13 notes · View notes
corre1310-blog · 6 months
Text
Tumblr media Tumblr media Tumblr media Tumblr media
Edenians Royalties if they're Muslim ruler
It feels like they're live in Turkey in Ottoman Era
39 notes · View notes
david-of-silence · 1 year
Text
Tumblr media Tumblr media Tumblr media
Bella Thorne
10 notes · View notes
hombresguapos · 5 days
Text
Tumblr media
2 notes · View notes
unrealityliminal · 9 months
Text
Tumblr media
3 notes · View notes