Tensor.Art: Configuring image generation settings in Tensor Art

Tensor.Art Try it →Watch full video →Artificial Quotient · Jul 2024

Demo summary

The video demonstrates how to input prompts and negative prompts while adjusting technical parameters like LoRA, ControlNet, VAE, aspect ratio, sampling methods, and CFG scale.

Step-by-step

Click the blue Create button
Enter your image description in the prompt column
Fill in the negative prompt column with elements to exclude, such as 'extra limbs'
Select an aspect ratio like portrait, landscape, or square
Adjust the Sampling steps to determine the refinement level
Set the CFG scale to control prompt adherence
Toggle on Upscale, Layer diffusion, or A Detailer if needed
Click to generate the image

Options

Add LoRA for specialized model styles
Use ControlNet with a reference image
Add embeddings
Manually select a VAE or leave it on automatic
Set a custom aspect ratio size
Change the Sampling method to alter the refinement approach
Input a specific Seed number to replicate a previous image
Enable Layer diffusion for transparent backgrounds

Watch out for

Higher sampling steps burn through credits faster
Higher quality settings consume more credits
ControlNet requires high-quality, clear reference images to be effective

Tips

Use the A Detailer to reduce face collapse and hand deformities
Experiment with VAE settings to see different color performances
Use a CFG scale of 7 as a starting point
Reuse a seed number from a successful generation to maintain consistency in your next image
Wait to explore advanced settings until you have a grasp of the basic parameters

All demos from “How To Use Tensor Art To Make Consistent Characters! (FREE Midjourney Alternative)”

0:440:34Navigating the Tensor Art interfaceThe narrator walks through the Tensor Art dashboard, showing the vertical tabs for AI tools, the Models tab, and the community leaderboard.Tensor.Art· AI Image Generator
2:002:31Configuring image generation settings in Tensor ArtCurrentThe video demonstrates how to input prompts and negative prompts while adjusting technical parameters like LoRA, ControlNet, VAE, aspect ratio, sampling methods, and CFG scale.Tensor.Art· AI Image Generator
4:410:43Generating an image from a promptThe user demonstrates the generation process and the final output of a high-quality image based on a detailed prompt about a boy skating.Tensor.Art· Text to Image
5:240:35Testing character consistency in Tensor ArtThe narrator demonstrates how to maintain character consistency by slightly modifying a prompt to change a character's pose while keeping their visual identity.Tensor.Art· AI Person Generator
Watch “How To Use Tensor Art To Make Consistent Characters! (FREE Midjourney Alternative)” →

AI Image Generator

0:140:44Upload a dataset to Tensor ArtThe user demonstrates how to navigate to the online training section of Tensor Art and upload a dataset of images for LoRA training.Bailey Chiu
1:250:46Select base model in Tensor ArtThe user shows how to choose a model theme and select a specific custom base model, such as Flux or Pony, for the training process.Bailey Chiu
0:440:34Navigating the Tensor Art interfaceThe narrator walks through the Tensor Art dashboard, showing the vertical tabs for AI tools, the Models tab, and the community leaderboard.Artificial Quotient
2:002:31Configuring image generation settings in Tensor ArtCurrentThe video demonstrates how to input prompts and negative prompts while adjusting technical parameters like LoRA, ControlNet, VAE, aspect ratio, sampling methods, and CFG scale.Artificial Quotient
0:201:40Applying LoRA and generating images in Tensor.artThe video demonstrates navigating to the Tensor.art generation workspace, selecting a base model, adding a LoRA style, entering a text prompt, and clicking generate to produce the final image.Harry Helps You