Nun sag’, wie hast du’s mit der #KI?
In der Redaktion setzen wir auf #Bildgeneratoren. In Bilddatenbanken finden wir kein passendes Bild zu Fällen wie „Diebstahl an der Wursttheke“ oder „Bienenwachsfontäne“.
Aber Probleme bei der Nutzung von #TextzuBildGeneratoren sorgen in unserer Redaktion immer wieder für Diskussionen. Von #Algorithmus & #geistigesEigentum bis #Umweltschutz.
Was meint ihr? Sollten wir die Finger davon lassen?
RT @AlexHinojo: New #Texttoimage research by @GoogleDeepMind proposes a benchmark that assesses the cultural competence of these models al…
via https://twitter.com/WikiResearch/status/1818807279694877159
https://www.makeuseof.com/what-is-google-imagen-ai
Google has its own text to image model now called Imagen.
It supports most of the advanced features people expect these says such as inpainting, outpainting, and using reference mages.
The article linked tells goes into a lot more detail about the project, and how you can start using it for free.
Addendae 1
Key-Locked Rank One Editing for Text-to-Image Personalization
https://research.nvidia.com/labs/par/Perfusion
https://news.ycombinator.com/item?id=37020620
* [NVIDIA] Perfusion: text-to-image method.
* 100KB model size/concept (excl. pretrained model: several GB), trained for ~4 min
* creatively portray personalized objects using novel mechanism: “Key-Locking”
Key-Locked Rank One Editing for Text-to-Image Personalization
https://arxiv.org/abs/2305.01644
The AI Engine that Fits in 100K - Running your own AI models is possible, but it requires a giant computer, right? M... - https://hackaday.com/2023/08/02/the-ai-engine-that-fits-in-100k/ #artificialintelligence #texttoimage #aimodel #ai
Si sta arrivando alla modifica di immagini attraverso l'#AI generativa (da #prompt testuale) che mantiene il contesto dettagliato.
Salesforce AI, ha presentato EDICT, una tecnica che sfrutta alcune caratteristiche dei modelli di diffusione per modificare solo alcuni dettagli dell'immagine originale.
Nelle immagini si vedono i risultati. 
Il paper: https://arxiv.org/pdf/2211.12446.pdf
Diffusion model, https://en.wikipedia.org/wiki/Diffusion_model
What are Diffusion Models? https://www.youtube.com/watch?v=fbLgFrlTnGU
For general introduction, these are excellent
* Stable Diffusion, https://en.wikipedia.org/wiki/Diffusion_model
* Stable Diffusion Got Supercharged - For Free! https://www.youtube.com/watch?v=1RvZWHtFXuY
* Stable Diffusion Is Getting Outrageously Good! https://www.youtube.com/watch?v=rzKVInXep5M
Midjourney V5 definitely got “handier”. (Couldn’t resist the pum.) While it is still far away from creating perfect hands, the results are way less awkward than in past versions. 
#AI #Midjourney #TextToImage #GenerativeAI
@AdamMackay Wow, superb comparison. I'll share that with my student, Ryan, who created the original. So amazing! THANKS! I definitely need to try Midjourney... #creativity #AI #texttoimage #edutooter #edtechSR #MediaLit
Stable Diffusion 2.0 Release, https://stability.ai/blog/stable-diffusion-v2-release.
The Stable Diffusion 2.0 release includes robust text-to-image models trained using a brand new text encoder (OpenCLIP), developed by LAION with support from Stability AI, which greatly improves the quality of the generated images compared to earlier V1 releases.
Repository: https://github.com/Stability-AI/stablediffusion
Tag 3 des AI-Experiments ist angebrochen und ich sage nur "Welcome to the Planet of the Apes" 
Bin gespannt ob wir von den Weltraum-Affen jemals wieder runterkommen. 4 Runden gibt es ja noch zu spielen!
Habe uebrigens auch noch einen Tumblr dazu gemacht um die einzelnen Schritte noch einmal schoen zu sammeln > https://www.tumblr.com/maiaiwai
Bevor ich morgen den naechsten Captain Future durch die AI-Muehle schiebe... Ihr duerft alle Bilder so nutzen wie ihr moechtet. Bitte bedient euch... genau dafuer ist das auch gedacht.
Und dies ist die erste "Image to Image" Version davon.
Wohlgemerkt mit diesen Parametern:
high definition, insanely detailed , Cartoon
Der Captain guckt schon ein wenig grimmiger :)
P.S. Morgen geht es weiter... Ich freue mich drauf!
Ich frickle schon seit Wochen mit den diversen "text to image" AIs rum und vor allen Dingen damit, was diese aus einem bereits bestehenden Bild machen.
Deshalb starte ich mal eine neue Reihe: 1 Bild pro Tag und was die AI dann innerhalb einer Woche draus macht.
Und es sind dabei immer die gleichen Parameter:
high definition, insanely detailed , Cartoon
Wir fangen mit einer Captain. Future Variante an.. das wird ein Spass!
'/%3e%3cpath%20d='M50.3943%2022.237V39.5876H43.5185V22.7481C43.5185%2019.2029%2042.0411%2017.3949%2039.036%2017.3949C35.7325%2017.3949%2034.0778%2019.5338%2034.0778%2023.7585V32.9759H27.2434V23.7585C27.2434%2019.5338%2025.5857%2017.3949%2022.2822%2017.3949C19.2949%2017.3949%2017.8027%2019.2029%2017.8027%2022.7481V39.5876H10.9298V22.237C10.9298%2018.6918%2011.835%2015.8754%2013.6453%2013.7877C15.5128%2011.7049%2017.9623%2010.6355%2021.0028%2010.6355C24.522%2010.6355%2027.1813%2011.9885%2028.9542%2014.6917L30.665%2017.5633L32.3788%2014.6917C34.1517%2011.9885%2036.811%2010.6355%2040.3243%2010.6355C43.3619%2010.6355%2045.8114%2011.7049%2047.6847%2013.7877C49.4931%2015.8734%2050.3963%2018.6899%2050.3943%2022.237Z'%20fill='white'/%3e%3cdefs%3e%3clinearGradient%20id='paint0_linear_89_8'%20x1='30.5'%20y1='0'%20x2='30.5'%20y2='65'%20gradientUnits='userSpaceOnUse'%3e%3cstop%20stop-color='%236364FF'/%3e%3cstop%20offset='1'%20stop-color='%23563ACC'/%3e%3c/linearGradient%3e%3c/defs%3e%3c/svg%3e)
