Each control type reads something different from your picture: its edges, its depth, a pose, its colours. This page shows what each one keeps and what it lets the model change. Press any control type to see it on three pictures, and to see what changes as you raise the control strength. Each of those comes as a character, a building and a flat colour drawing, so you can see what a detector makes of a face, of straight edges and of flat colour.
The image you get back is not a copy of your picture. A detector reads your picture first and draws a control map from it: outlines, a clean line drawing, how far away each part is, which way each surface faces, where the joints of a body are, or flat colour fields. The worker then makes a new image that follows that map. So choosing a control type is choosing what has to survive from your picture.
Each control type here is shown on three pictures, because what a detector keeps depends on what it is given. Press one to see all three. For each you get the picture it started from, the control map the detector drew, and the image that map guided. Everything else is the same across all of them: one model, one prompt, one seed, one control strength.
Edges and outlines
These read the outlines of your picture. The image keeps the same shapes, and the colour, the material and the style can all change.
canny Finds hard outlines a pixel wide wherever light meets dark. It makes no distinction between the edge of an object and the edge of a texture, so a detailed photograph gives a busy map.
hed Reads soft edges, following the shapes a person would trace rather than every hard boundary, which suits a photograph better than canny does.
pidinet Reads soft edges like hed, thinner and cleaner, and runs quicker.
teed The newest of the soft edge readers, with thin continuous lines and less speckle than hed or pidinet.
pyracanny Runs canny over several sizes of your picture at once, so the large shapes survive and the fine texture drops out.
binary Turns everything darker than a threshold black and the rest white, with no edge finding at all, so it wants a stark drawing to start from.
scribble Traces the dark parts of your picture coarsely, on purpose, and expects a drawing on a pale ground; on a photograph use fakescribbles.
fakescribbles Thins the soft edge map down to a loose hand-drawn scribble, which is the one to use on a photograph.
scribble_xdog Draws a scribble from the difference between two blurs, with finer strokes than the plain scribble and more of the faint detail kept.
scribble_pidinet Thins pidinet's edges to a scribble, cleaner and more continuous than the other scribbles.
Line art and straight lines
These draw your picture as clean lines. The ones that keep straight lines only leave you the architecture and drop everything curved.
standard_lineart Calculates a plain ink drawing directly from the pixels, with no model involved. It is fast, and it loses faint detail.
lineart Traces a photograph the way an illustrator would ink it, with weighted strokes and the faintest detail left out.
lineart_anime Draws the lines in the anime idiom, with even strokes and closed shapes ready to be coloured in.
lineart_anime_denoise Draws anime lines after cleaning out paper grain and screentone, which suits scans and heavily compressed pictures.
mlsd Keeps straight line segments only and discards everything curved, which usually leaves the architecture. hough is the older spelling of this detector, and the horde still accepts it. Older workers understand that spelling.
Depth
These read how far each part of your picture is from the camera. The image keeps the same arrangement in space, and the drawing itself can change.
depth Reads how far away each part of your picture is. It is reliable across a whole scene, though it loses small objects.
midas_depth The older and faster depth map, coarser than the rest and steady on an ordinary photograph.
zoe_depth Reads depth in real distances rather than relative ones, so near and far stay in proportion across the picture.
depth_anything A newer depth map that holds fine detail: hair, railings, leaves.
depth_anything_v2 The sharpest depth map on offer, with cleaner boundaries between objects and less smearing than depth_anything.
Surface direction
These read which way each surface faces. The image keeps the same modelling and light, and the model redraws the picture.
normal Works out which way each surface faces from a depth map, rougher than normal_bae and tolerant of an odd picture.
normal_bae Reads which way each surface faces with a model trained for it, the cleaner of the two and the usual pick.
Pose
These read the people and nothing else. The same stance comes back in any setting.
openpose Draws the people as skeletons, with body, hands and face as coloured sticks and dots on black, and discards everything else.
Regions and objects
These read your picture as flat fields, one for each kind of thing. The image keeps the same layout, and every surface is redrawn.
seg Splits your picture into flat coloured fields, one colour for each kind of thing, in the quick and coarse version.
oneformer_ade20k The same split done by a stronger model over a scenery vocabulary: walls, sky, floor, furniture, plants.
oneformer_coco The same split over an object vocabulary, which names people, animals and things carried or sat on rather than the parts of a room.
Colour and texture
These read the palette or the tones. The tile readers read the picture itself, which is how you add detail to it or enlarge it under guidance.
color Reduces your picture to a grid of flat colour blocks, keeping the palette and where each colour sits, and nothing of the drawing.
shuffle Distorts your picture into a swirl. The colours and textures are kept, and the composition is destroyed. Use it to pass on a picture's palette and feel without its layout.
recolor_luminance Makes a greyscale copy, weighted the way the human eye judges brightness. Use it as the map when you want to change a picture's colours but keep its shading.
recolor_intensity Makes a grey copy from the strongest colour channel, which leaves saturated colours brighter than the luminance version does.
tile Blurs your picture to its broad shapes, which is what a tile ControlNet needs to add detail without inventing a different picture.
tile_ttplanet_guided A gentler tile map that blurs the flat areas and leaves the edges crisp, and the steadier one on a photograph.
tile_ttplanet_simple The plain tile map, an even blur with no edges kept, quicker and enough when your picture is already clean.