Monday 21 November 2022

November 20, 2022

 I drew at the Edith Bunker in Albuquerque:

AI variations of the long pose


Drawn on Stipple Paper







AI VARIATIONS

Image plus Midjourney prompt: "in the style of Pontormo"
Initial generation
 plus second variations generation




Image plus Midjourney prompt: "in the style of origami"
I mixed the first two image JPGs in Midjourney
to get the four options in the last image

I am not that excited
with this result



Image plus Midjourney prompt: "designed in Pepakura"




Image plus Midjourney prompt: "two heads"





Image plus Midjourney prompt: "in the style of Op Art"





Image plus Midjourney prompt: "in the style of Roberto Matta"




Image plus
 Midjourney prompt: "3D model created in SolidWorks"



Image plus Midjourney prompt: "in the style of Hans Bellmer"




Wrapped Experiment




Since I often try to add energy orbiting around the figure, I used the prompt: "woman wrapped in cords" to get Midjourney to add that kind of surround static around a female figure.

1)  First I just entered the prompt: "woman wrapped in cords" and nothing else, without uploading a figure.  Midjourney always seems to default to a portrait:


Midjourney prompt only


2)  Then I uploaded one of my figures, followed by the prompt: "woman wrapped in cords," and generated the following:

My drawing plus the Midjourney prompt


I liked where this result is going


3)  Then I uploaded my more abstract figure drawing, with more energy, and again the same prompt: "woman wrapped in cords."  I think the results generated had more energy, and thus were a little more successful:




My more abstract drawing
plus the Midjourney prompt


4)  Then I mixed images --my more abstract drawing with the most successful result -- to see if I could generate a figure that was more in my style:



My drawing uploaded
plus a previously generated Midjourney image
from the same drawing

My drawing does influence the Midjourney results, but it would be nice to have a little more personal sway over the image generated.

  • Midjourney seems to add the same default woman's head when it can squeeze that in there.  And also the same female proportions.  
  • Robert recommends tagging on "--chaos 10" for less variation, or "--chaos 90" for more extreme variations. 

5) I kept reworking it.  I uploaded the previous medicore result adding the same prompt: "woman wrapped in cords," and generated some better images:








Other:

No comments:

Post a Comment

Note: only a member of this blog may post a comment.