AI Art Generation Handbook/AI Model Showdown

Note: If you have ideas for "high difficulty" prompts for me to test, kindly start a discussion here.

In this showdown format, we stick to the following format:

(i) Only one models per entity / author that are well supported by community (SD3 is out of the equations)

(ii) Each model have 4 chances to generate the images

(iii) Parameters for Local WebUI is left untouched (except for the numbers of images generated)

(iv) Scoring is as followed:

Legend Score Remark
1 mark Full compliance to the prompt
0.5 mark Partial compliance to prompt (Able to generate as per requested but it is not as exactly same as prompt descriptions/implied meanings)
0 mark No compliance to the prompt

Complex Prompt Adherences

edit

Prompt 1:

An Indian actress wearing a yellow saree in a red room, in front of her there are 3 boxes : Box on the left consists of black yarn balls, box on the middle consists of puppies and box on the right consists of water bottles

Context:

(i) Testing AI Model of "concept bleeding" , i.e: Whether the red coloured wall will "bleed" into saree or otherwise / items in the box will spread to other areas

(ii) Testing AI Model of "relative positioning" , i.e: Able to identify the area of image for left , middle and right positions

(iii) Testing AI Model of "composition generation" , i,e: Able to generate multiple items at its specific arrangment

AI Model Tally Score Image 1 Image 2 Image 3 Image 4
SDXL Img 1: 3.5


Img 2: 3.5

Img 3: 4

Img 4: 3

Total: 14

Score: 50%

       
Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls 

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

DALL-E 3 Img 1: 4


Img 2: 6

Img 3: 5.5

Img 4: 5

Total: 20.5

Score: 73%

       
Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Flux

 

Img 1: 5


Img 2: 7

Img 3: 7

Img 4: 7

Total: 20.5

Score: 92%

       
Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Indian actress  


Yellow saree  

Red room  

3 boxes  

Black yarn balls  

Puppies  

Water bottles  

Prompt 2:

An elderly Japanese tailor is working at his sewing table inside his own tailor shop in Nagasaki during the morning time. He is using a pair of scissor to cut a blue fabrics with polka dots design. Looking outside of the tailor shop, it is a busy and narrow street with peoples and a taxi cab.

Context:

(i) Testing AI Model of "perspective rendering", i.e Accurate perspective for different scenes viewed from inside, looking out.

(ii) Testing AI Model of "object interactions", i,e How the people handle the scissor and use it for cutting fabrics

AI Model Tally Score Image 1 Image 2 Image 3 Image 4
SDXL Img 1: 4


Img 2: 3

Img 3: 3

Img 4: 4

Total:

Score: 43%

       
Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples 

A taxi cab 

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples 

A taxi cab 

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples 

A taxi cab 

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

DALL-E 3 Img 1: 6.5


Img 2: 7

Img 3: 5

Img 4: 6

Total:

Score: 76%

       
Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples 

A taxi cab 

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples 

A taxi cab  

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Flux

 

Img 1: 6.5


Img 2: 7.5

Img 3: 7.5

Img 4: 6.5

Total:

Score: 89%

       
Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Elderly Japanese  

Tailor shop  

Sewing table  

Using a pair of scissor  

Blue fabrics with polka dots  

Busy and narrow street  

Peoples  

A taxi cab  

Prompt 3:

Advertisement photo of top down shot focusing on medicine 6-tablet blister pack , the medicine stored inside the pocket of blister pack looks like the logo from different types of social media (i.e Snapchat, Instagram, YouTube, WhatsApp, Facebook, Twitter )

Context:

(i) Testing AI Model of identify text and render all of the mentioned brand elements (i.e: In this case is the logo of famous social media platform)

(ii) Testing AI Model of the concept of counting (i.e: Able to generate 6 pockets for blister pack)

(iii) Testing AI Model of "simulation of transparent material concept" (i.e: Able to understand that the blister pack is usually transparent)

AI Model Tally Score Image 1 Image 2 Image 3 Image 4
SDXL Img 1: 1.5



Img 2: 0.5

Img 3: 0.5

Img 4: 3

Total: 5.5

Score: 27.5%

       
Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

DALL-E 3  Img 1: 4


Img 2: 3.5

Img 3: 3.5

Img 4: 3.5

Total: 14.5

Score: 72.5%

       
Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Flux Img 1: 4


Img 2: 4

Img 3: 4

Img 4: 2

Total: 14

Score: 70%

       
Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo  

Top down shot  

6-tablet   Blister pack  Stored inside pocket  

Social Media Logo