Dezgo AI : BlowFishMan and friends

See previous Text to Image blog https://prataverse.blogspot.com/2023/07/dazgo-ai-text-to-image-lizard-man.html

BlowFishMan

Text to Image / Vintedois Diffusion Model.

PROMPT : Blowfishman

ClamShellMan

Text to Image / Vintedois Diffusion Model.

PROMPT : Clam man

 

FishHeadMan

Text to Image / Vintedois Diffusion Model.

PROMPT : Fishheadman

 

Prompt : Piranha fish head man.

Photoshop Generative fill : lasso the tongue and click generate to remove it.

It is not the fish I have in mind. 

Photoshop Generative fill : lasso the fish.

Prompt : Piranha fish

It is not a piranha fish but I do like the facial expression.

 Photoshop Generative fill : lasso the body of the fish but not the head.

The skin color of man is not the same as fish.

Select the fish , adjust the hue and saturation to the right color . 

Next use the blur tool to blur the area of fish connected to the neck. 

Finally use the burn tool to darken the fish , the area near the neck.

 Piranha Fish Head Man

Controlled Text to Image / Vintedois Diffusion Model.

PROMPT : Pirana 

This is generated using Controlled Text to Image feature. To see more https://prataverse.blogspot.com/2023/07/dezgo-controlled-text-to-image.html

 SharkMan

Instead of changing the man's head into a shark, I doodle the shark walking upright like a man in fine leather boots.

The extra line works on the shark would pose a problem for Controlled text to image to process. See https://prataverse.blogspot.com/2023/07/dezgo-controlled-text-to-image.html So for this case Text to image option seem to be a better choice.

Text to Image / Vintedois Diffusion Model.

PROMPT : Shark walking like a man

 

Dazgo AI : Image to Image Generation

 Instructional Doodling Process

 Refer to https://prataverse.blogspot.com/2023/04/instructional-doodling-introduction.html for my definition of instructional doodling after all it is the oxymoronic phrase that I have invented. 

I do frequent doodling on my phone. I call the process instructional doodling where like AI image generator I give myself a "prompt" or instruction for the doodling. Usually the process is as follows...

• STEP 1 : Just select a color and size of brush. Doodle mindlessly and freely to get to my 1st doodle image.
  
  Change to another color and brush size  if nothing comes out from the doodle.
  
  If my mind is preoccupied then nothing good will come out of the doodle.
  
  If I have something in mind and I am trying to replicate the image from my mind then I am drawing , not doodling.
  
  The doodling state of mind is subconscious , I don't really know what I am going to get. If my mind is full of thoughts I cannot doodle.
  
  
• STEP 2 : After getting the 1st doodle image and somewhat satisfied with its direction then create a general instruction for a series of doodling. This will create a focus and direction to my doodles at this moment in time.
  
  
• STEP 3 : Create more doodles according to the instruction. These series should be continuous , do not take a break or do it another day. Do the whole series until I hit a brick wall and then I stop for the day.

Now with Image to Image feature , 

• STEP 4 : Start step 4 only after completed all doodles. See how AI can reinterpret each doodle image.

 

 

INSTRUCTION ( Step 2 ) : Stick figure of bodies.

PROMPT ( Step 4 ) : Cat with elongated body. 

Adjust the Strength , the more the percentage, it the more it will deviate from  the image.

 

The result is impressive, a six legged cat , maybe an extra two legs to support the longish body.  

INSTRUCTION ( Step 2 ) : Stick figure of bodies.

 PROMPT ( Step 4 ) : Two dancers

 

INSTRUCTION ( Step 2 ) : Stick figure of bodies.

 PROMPT ( Step 4 ) : Two acrobats hands holding legs.

 

INSTRUCTION ( Step 2 ) : 3D sculptural line work

 PROMPT ( Step 4 ) : Marble sculpture face.

 

 
 PROMPT ( Step 4 ) : Marble sculpture .

 

 

INSTRUCTION ( Step 2 ) : Imagine organs inside my body.

 

PROMPT ( Step 4 ) : Intestines realistic rendering

 

INSTRUCTION ( Step 2 ) : Imagine organs inside my body.

PROMPT ( Step 4 ) : Cell human anatomy realistic rendering

INSTRUCTION ( Step 2 ) : Imagine organs inside my body.

PROMPT ( Step 4 ) : Brain and skull cell anatomy realistic rendering

See the next Image to Image blog https://prataverse.blogspot.com/2023/07/dezgo-ai-cubist-to-realist.html

Dazgo AI : Text to Image Lizard Man Character

See previous Text to Image Blog https://prataverse.blogspot.com/2023/07/dezgo-ai-random-generation-text-to-image.html

Dazgo AI has an excellent generation for the Lizard Man Character using the AbsoluteReality Model. I accessed the Dezgo website using my android phone.

 

LIZARDMAN

Prompt: Lizard man

 Prompt : Lizard man half body shot.

 Prompt : Lizard man squatting and flexing muscles.

Prompt : Lizard man in water flexing muscles full body shot

 Prompt : Lizard man flexing muscles in fire background

Prompt : Lizard man flexing muscles in the lightning storm.

Prompt : Wet Lizard man flexing muscles in heavy rain.

In Adobe Photoshop Beta , I extend the canvas to 640 px by 640px. The blank space is selected and click on the generate button of the generative fill.

 The body is extended quite decently.

LizardWoman

Prompt : Lizardwoman ballet dancer full body shot.

I have generated quite many images to get a decent version of lizard woman.

 

Edit Image with Instructions . It is not possible to extend the tail using this so I have decided to change the background.

Prompt : Add snow

After that I use Adobe Photoshop Beta to lasso the tail and hand area to run the AI generative fill. 

Prompt : Lizardwoman dancing ballet.

Adobe Photoshop Beta Prompt : Extend lizard leg

After several edits the results looks good.

See the next Text to Image blog https://prataverse.blogspot.com/2023/07/dezgo-ai-blowfishman-and-friends.html

Dezgo AI : Random Generation Text-to-Image

This is a new Image Generator besides the Adobe Firefly beta that I have tested. https://prataverse.blogspot.com/2023/06/adobe-firefly-beta-abstract-painting.html

I am surprised that the image generation is great in quality , no general distortion of faces and hands. It generates one image at one time. I use model DreamShaper 6 for the following images. I used an android phone.

Prompt: Superman Flying with arms in front and clenched fist.  

Click the Run button and wait for the image to generate. It can take a few minutes to generate.

I am not sure if this generated image of Superman infringes on the trademark. It is just a test if AI can generate brands or icons of popular culture.
 

Click Download to download the picture.    

Prompt : Cat angel

It generate nice cute picture. 

 

 

Prompt: Chair with lights

It can combine two ideas into a product design.

Prompt : UFO landing on planet Mars.

Render good science fiction environment.
 

Prompt : Tokyo Yakuza gangster with dragon tattoo on his back.

Quite good result for the picture "dragon" on another picture "gangster back". Another example is graphic design on a T-shirt or a logo on an aeroplane.

This is my favorite happy accident image. 

I wanted to create an image of a man eating an octopus. If I prompt man eating food , the AI generates a normal picture but once I swap the word food with octopus , the generation gets bizarre.

Prompt : Close up view of Japanese man bites the octopus.

Prompt : Alien nation.

Prompt : Japanese tori gate in the middle of the ocean with choppy waves

See my next Text to Image Blog https://prataverse.blogspot.com/2023/07/dazgo-ai-text-to-image-lizard-man.html