← All courses ← Course home

Lesson 03 of 06

Diffusion: make images and video

The diffusion type is for making new pictures or video. Use it when a prompt should become a new image or a new video.

Agent brief (llms.md)

A diffusion app starts with a description and makes media. That is a different shape from a classifier. A classifier answers what is already in a picture. A diffusion model makes a picture from a prompt, or in Wan's case a video.

InputText prompt
diffusionNew-media preset
PromptDescribe the scene
ModelMake and refine media
OutputNew pixels or frames

Stable Diffusion family

The catalog lists sd-1.5, sd-2.1, and sd-3.5-medium. They are Stable Diffusion family presets for text-to-image work. If the feature is a prompt that should create an image, this is the first family group to look at.

FLUX.2 Klein

flux2-klein-4b is the catalog's FLUX.2 Klein preset. Its listed job is also text to image. Keep this clear: this is still the path for making a new image, even though the family name is different from Stable Diffusion.

Wan

wan-t2v-1.3b is the Wan text-to-video preset. Choose this when the output is moving frames, not one still image. The job changes from prompt-to-image into prompt-to-video.

PromptDescribe what to make
Stillsd or flux2-klein-4b
Movingwan-t2v-1.3b

The image-model lesson shows the broader shape difference. The ClothingCNN course shows a model that labels an existing image, not one that draws a new one. From the metal to the model follows the chosen generation recipe into the work the chip actually runs.

Use diffusion when the app must create media. If it only needs to say what is in an existing image, use a utility vision type instead.

Next, look at those utility vision jobs, from labels to cut-outs and depth.

Key concepts

  • diffusion turns a prompt into new media, not a label for an existing image.
  • Still-image presets include sd-1.5, sd-2.1, sd-3.5-medium, and flux2-klein-4b.
  • wan-t2v-1.3b is the Wan text-to-video preset.
  • Classification describes what is already in a picture. Diffusion draws new pixels or frames.

Takeaways

  • You can pick diffusion when the app must create media from a prompt.
  • You can choose a still-image preset or wan-t2v-1.3b for moving frames.
  • You can say when utility vision fits better than diffusion for an existing photo.