Was Stable Diffusion so über Velomobile denkt:

Es fiel mir ja anfangs schwer gut funktionierende Prompts für Stable Diffusion zu finden. Zwischenzeitlich hatte ich Im Internet oder vielleicht sogar hier im Thread gelesen, dass es sogar eigens Programme für das Optimieren der Prompts gäbe. Ich habe jetzt mal jegliche Intelligenz ausgelagert und einfach ganz dumm ChatGPT gefragt. Auf Englisch, um nicht auch noch eine Übersetzungs-KI dabei zu haben. Ein sonntäglicher Trialog zwischen einer HI und zwei AI:

HICould you tell me a prompt for Stable Diffusion to generate images with recumbent trikes?
Text-AISure, here's a prompt you can use for Stable Diffusion to generate images with recumbent trikes:
"Generate a series of 10 images featuring recumbent trikes in different settings using Stable Diffusion. Each image should showcase a recumbent trike in a different environment, such as a forest, beach, city street, or mountain trail. The trikes should be the focal point of the image, and the background should be blurred to create a sense of motion. Use warm colors to create a sense of excitement and energy."
You can adjust the number of images and settings as per your preference. Don't forget to experiment with different parameters to get the desired output. Good luck!
Bild-AI20230226104302_2805353717.png
HINo, I just wanted to have one trike per picture. Could you also do something to cheer up the images? Something like blue skies?
Text-AICertainly! Here's an updated prompt for Stable Diffusion to generate images of recumbent trikes in different settings with a cheerful and vibrant feel:
"Generate a series of 10 images featuring recumbent trikes in different settings using Stable Diffusion. Each image should showcase a single recumbent trike in a unique and picturesque location such as a beach, park, or countryside. The trike should be positioned in a way that draws attention to its sleek design and unique features. Use vibrant colors and incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Experiment with different blur levels and lighting effects to create a dynamic and visually appealing image. Each image should be a celebration of the freedom and joy of cycling."
Feel free to modify this prompt as per your preferences, and don't forget to adjust the parameters in Stable Diffusion to achieve the desired output. Have fun generating your images!
HII see. Leave away the instructions about the 10 pictures.
Text-AISure, here's an updated prompt without the instruction to generate 10 images:
"Using Stable Diffusion, create a vibrant and visually appealing image featuring a single recumbent trike in a picturesque setting such as a beach, park, or countryside. The trike should be positioned in a way that showcases its sleek design and unique features. Incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Use vibrant colors and experiment with different blur levels and lighting effects to create a dynamic and cheerful image that celebrates the freedom and joy of cycling."
Feel free to modify this prompt as per your preferences and adjust the parameters in Stable Diffusion to achieve the desired output. Have fun creating your image!
Bild-AI20230226104730_465286128.png
HIStill the trikes are far too strechted and most importantly the number of wheels don't match. Could you try again for a tadpole recumbent trike that exactly has three wheels, which is two wheels in the front and one wheel at the back. Thanks.
Text-AII apologize for the confusion in my previous responses. Here's an updated prompt for Stable Diffusion to generate an image of a tadpole recumbent trike with three wheels:
"Using Stable Diffusion, create a visually appealing image of a tadpole recumbent trike with three wheels, positioned in a picturesque setting such as a beach, park, or countryside. The trike should be positioned in a way that showcases its sleek design and unique features, with the two front wheels prominently visible and the rear wheel slightly blurred. Incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Use vibrant colors and experiment with different blur levels and lighting effects to create a dynamic and cheerful image that celebrates the freedom and joy of cycling on a recumbent trike."
Again, feel free to modify this prompt as per your preferences and adjust the parameters in Stable Diffusion to achieve the desired output. I hope this updated prompt helps you create the image you're looking for!
Bild-AI20230226105129_1812026785.png
HIYeah, I think your instruction for Stable Diffusion is quite precise, but Stable Diffusion still does not understand correctly. What could we do to make Stable Diffusion really understand your prompt?
Text-AII'm sorry to hear that the previous prompts didn't work as intended. Here are a few tips to help ensure that Stable Diffusion understands your prompt:
  1. Be specific: Try to be as specific as possible with your prompt, including details such as the exact number of wheels, the positioning of the trike, and the specific setting you want to use.
  2. Use examples: Including examples of the type of image you're looking for can help Stable Diffusion better understand your prompt. You can include links to images or describe them in detail.
  3. Experiment with parameters: Stable Diffusion has several parameters that can be adjusted to achieve the desired output. Experimenting with different values for these parameters can help fine-tune the results.
  4. Keep it simple: Sometimes, simpler prompts can yield better results. Instead of trying to include too many details, focus on the key elements that are important to you.
I hope these tips help you create the image you're looking for using Stable Diffusion. Good luck!
HIThanks a lot for your help and your time today. See you.
Text-AIYou're welcome! It was my pleasure to help. Don't hesitate to reach out if you have any more questions in the future. Take care!

Vielleicht hätte ich noch weiter machen sollen, indem ich ChatGPT auf dessen eigenen Punkt 4 hinwies. Oder doch noch die Übersetzungs-AI mit ins Boot holen? Das gibt aber insgesamt einen Eindruck, wie Entwicklungsprozesse zukünftig aussehen könnten. Parallel zur Bild-AI hätte man noch eine CAD-AI, die dann eine 3D-Kohlenstoffnanoröhrendrucker-AI füttert. Ich hätte gar nicht genug Garagen, um all die gedruckten Fahrräder unterzubringen. Man bräuchte auf jeden Fall auch eine Recycler-AI. Vielleicht habe ich aber auch nur zu viele Marvel-Filme geguckt. Apropos, bei Star Trek gibt es ja diese eine Szene mit Scotty und dem transparenten Aluminium. Zum Erholung geht ischt gloicht noch ein wenich Realradeln. Cheers. PS: Schreibfehler hat die Üversetzungs-AI zu verantworten.
 
a woman in a velomobile by spitzweg_1.jpga woman in a velomobile by spitzweg_2.jpga woman in a velomobile by spitzweg_3.jpga woman in a velomobile by spitzweg_4.jpg
"a woman in a velomobile painted by spitzweg".

Schon recht erstaunlich, wie gut die Fahrzeuge das damalige Zeitkolorit verkörpern.
 
Es fiel mir ja anfangs schwer gut funktionierende Prompts für Stable Diffusion zu finden. Zwischenzeitlich hatte ich Im Internet oder vielleicht sogar hier im Thread gelesen, dass es sogar eigens Programme für das Optimieren der Prompts gäbe. Ich habe jetzt mal jegliche Intelligenz ausgelagert und einfach ganz dumm ChatGPT gefragt. Auf Englisch, um nicht auch noch eine Übersetzungs-KI dabei zu haben. Ein sonntäglicher Trialog zwischen einer HI und zwei AI:

HICould you tell me a prompt for Stable Diffusion to generate images with recumbent trikes?
Text-AISure, here's a prompt you can use for Stable Diffusion to generate images with recumbent trikes:
"Generate a series of 10 images featuring recumbent trikes in different settings using Stable Diffusion. Each image should showcase a recumbent trike in a different environment, such as a forest, beach, city street, or mountain trail. The trikes should be the focal point of the image, and the background should be blurred to create a sense of motion. Use warm colors to create a sense of excitement and energy."
You can adjust the number of images and settings as per your preference. Don't forget to experiment with different parameters to get the desired output. Good luck!
Bild-AIAnhang anzeigen 317981
HINo, I just wanted to have one trike per picture. Could you also do something to cheer up the images? Something like blue skies?
Text-AICertainly! Here's an updated prompt for Stable Diffusion to generate images of recumbent trikes in different settings with a cheerful and vibrant feel:
"Generate a series of 10 images featuring recumbent trikes in different settings using Stable Diffusion. Each image should showcase a single recumbent trike in a unique and picturesque location such as a beach, park, or countryside. The trike should be positioned in a way that draws attention to its sleek design and unique features. Use vibrant colors and incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Experiment with different blur levels and lighting effects to create a dynamic and visually appealing image. Each image should be a celebration of the freedom and joy of cycling."
Feel free to modify this prompt as per your preferences, and don't forget to adjust the parameters in Stable Diffusion to achieve the desired output. Have fun generating your images!
HII see. Leave away the instructions about the 10 pictures.
Text-AISure, here's an updated prompt without the instruction to generate 10 images:
"Using Stable Diffusion, create a vibrant and visually appealing image featuring a single recumbent trike in a picturesque setting such as a beach, park, or countryside. The trike should be positioned in a way that showcases its sleek design and unique features. Incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Use vibrant colors and experiment with different blur levels and lighting effects to create a dynamic and cheerful image that celebrates the freedom and joy of cycling."
Feel free to modify this prompt as per your preferences and adjust the parameters in Stable Diffusion to achieve the desired output. Have fun creating your image!
Bild-AIAnhang anzeigen 317983
HIStill the trikes are far too strechted and most importantly the number of wheels don't match. Could you try again for a tadpole recumbent trike that exactly has three wheels, which is two wheels in the front and one wheel at the back. Thanks.
Text-AII apologize for the confusion in my previous responses. Here's an updated prompt for Stable Diffusion to generate an image of a tadpole recumbent trike with three wheels:
"Using Stable Diffusion, create a visually appealing image of a tadpole recumbent trike with three wheels, positioned in a picturesque setting such as a beach, park, or countryside. The trike should be positioned in a way that showcases its sleek design and unique features, with the two front wheels prominently visible and the rear wheel slightly blurred. Incorporate blue skies, fluffy clouds, and other elements that evoke a sense of joy and optimism. Use vibrant colors and experiment with different blur levels and lighting effects to create a dynamic and cheerful image that celebrates the freedom and joy of cycling on a recumbent trike."
Again, feel free to modify this prompt as per your preferences and adjust the parameters in Stable Diffusion to achieve the desired output. I hope this updated prompt helps you create the image you're looking for!
Bild-AIAnhang anzeigen 317984
HIYeah, I think your instruction for Stable Diffusion is quite precise, but Stable Diffusion still does not understand correctly. What could we do to make Stable Diffusion really understand your prompt?
Text-AII'm sorry to hear that the previous prompts didn't work as intended. Here are a few tips to help ensure that Stable Diffusion understands your prompt:
  1. Be specific: Try to be as specific as possible with your prompt, including details such as the exact number of wheels, the positioning of the trike, and the specific setting you want to use.
  2. Use examples: Including examples of the type of image you're looking for can help Stable Diffusion better understand your prompt. You can include links to images or describe them in detail.
  3. Experiment with parameters: Stable Diffusion has several parameters that can be adjusted to achieve the desired output. Experimenting with different values for these parameters can help fine-tune the results.
  4. Keep it simple: Sometimes, simpler prompts can yield better results. Instead of trying to include too many details, focus on the key elements that are important to you.
I hope these tips help you create the image you're looking for using Stable Diffusion. Good luck!
HIThanks a lot for your help and your time today. See you.
Text-AIYou're welcome! It was my pleasure to help. Don't hesitate to reach out if you have any more questions in the future. Take care!

Vielleicht hätte ich noch weiter machen sollen, indem ich ChatGPT auf dessen eigenen Punkt 4 hinwies. Oder doch noch die Übersetzungs-AI mit ins Boot holen? Das gibt aber insgesamt einen Eindruck, wie Entwicklungsprozesse zukünftig aussehen könnten. Parallel zur Bild-AI hätte man noch eine CAD-AI, die dann eine 3D-Kohlenstoffnanoröhrendrucker-AI füttert. Ich hätte gar nicht genug Garagen, um all die gedruckten Fahrräder unterzubringen. Man bräuchte auf jeden Fall auch eine Recycler-AI. Vielleicht habe ich aber auch nur zu viele Marvel-Filme geguckt. Apropos, bei Star Trek gibt es ja diese eine Szene mit Scotty und dem transparenten Aluminium. Zum Erholung geht ischt gloicht noch ein wenich Realradeln. Cheers. PS: Schreibfehler hat die Üversetzungs-AI zu verantworten.


ChatGPT nach StableDiffusion Prompts gefragt hab ich auch mal ... hab da aber Null Erfolg mit gehabt, mir scheint, dass ChatGPT tatsächlich absolut keine Ahnung hat wie man gute Prompts schreibt.
 
Prompt: "Ich hätte gerne eine Bleistiftzeichnung von einem Velomobil, zeichne ein geschlossenes, vollverkleidetes Liegefahrrad in aerodynamischer Form. Die Räder sollen von außen nicht sichtbar sein".

ChatGPT - DALL *E:

DALL·E 2023-11-15 16.20.06 - A pencil drawing of a closed, fully faired recumbent bicycle with...png

DALL·E 2023-11-15 16.40.36 - Create a photorealistic pencil drawing of a velomobile with a tap...png
 
Zuletzt bearbeitet:
Ich habe alles mögliche versucht, es wird kaum besser.... Auch nicht mit etlichen Bildvorlagen. Es kommen nur wilde Geschosse raus, die die Welt nicht braucht und die unfahrbar sind...

1700063344028.png

(Klugschwätzen kann das Ding jedenfalls.... !) :LOL:
 
Irgendwann werden die Ergebnisse besser werden. Und dann kann sowas schnell interessant sein. CAD-Dateien kann man übrigens anpassen...
 
Was macht das Ding übrigens, wenn man zusätzlich angibt, dass Holz in Stitch&Glue-Bauweise gewünscht ist
 
Im Vergleich zu den Abbildungen in Beitrag #61,62 schon einiges besser.
Man müßte die ki speziell mit Ergonomie, Fahrradtechnik und Aerodynamik füttern damit sie "versteht" was sie machen soll..
 
Zuletzt bearbeitet:
CAD-Dateien kann man übrigens anpassen...
Nein!?! Wirklich? Kann ja nicht sein...
Und was hilft das, wenn man nicht versteht, warum dort eine Kante zur Verstärkung an der einen Stelle eingefügt ist? Oder glaubst du wirklich, dass das Programm dir auch die entsprechende (und korrekte) Erklärung mitliefern wird?
 
KI-Antwort: "Hier sind noch zwei Bilder des Liegefahrrads mit der Kohlefaserhülle, ergänzt durch eine Haube mit Visierscheibe. Diese Konstruktion betont die Windschnittigkeit und das futuristische Design."

file-uLWfhg0ZZDSuSbrRsBgkzlh8.jpg

file-OuZRsKdaPcmPenqKdprF6Bcc.jpg
 
Man müßte die ki speziell mit Ergonomie, Fahrradtechnik und Aerodynamik füttern damit sie "versteht" was sie machen soll..
Dann kommt das dabei raus, was ein Daniel Fenn mit seiner Intelligenz entwirft, baut und verkauft.
KI, die neueste Sau, die durchs Dorf getrieben wird. Für die Doofen zum Staunen.
Oder für die, die selbst keinen geraden Satz aufs Papier bringen.
 
Zuletzt bearbeitet:
die beiden letzten Bilder von @BorisD schauen beeindruckend aus..
es erinnert mich - schon vor K.I. - was die letzten Jahre so in den "Tech-Medien" (oder auch weniger TEch) abgeht:

da wird wöchentlich irgendein neues Flugzeug vorgestellt, Überschall-Jet, 2-Personen-Uboot, 4-Personen-Drohne, ...
was weiß ich..
immer mit dem Satz, dass das Teil in +/-3 Jahren Marktreif sein soll...

und jedesmal sind die Artikel flankiert von einzig und allein 3D-DEsign-Studien, Bilder die teils von Praktikanten denen man nach 3 Monaten mal ne Aufgabe fürs Cad-Programm gibt, gerendert wurden
wo ich als Laie schon sehe, dass das doch Schmarrn ist..

vondaher: soviel anderes macht die K.I. wohl doch gar ned
 
Zurück
Oben Unten