Every pose in the library comes with two sets of images: one depth map and one faceless line art. They are two representations of the same pose, their filenames correspond one to one, and you can switch between them or compare them side by side at any time. This is the biggest difference between this library and an ordinary pose image site.
When to use which set
Depth map: fixes the skeleton
A depth map records how far each pixel is from the lens. Near is bright, far is dark, and the figure’s orientation, limb positions and front-to-back occlusion are compressed into a “distance map”.
Its value is this: the model is not dragged around by the source image’s materials, colours or lighting, but it does inherit the skeleton accurately. Drop it into ControlNet’s Depth channel and the pose holds. For the specific parameters (weight from 0.65, control end 0 to 0.75, preprocessor set to none) see How to Use ControlNet Depth Maps.
Faceless line art: fixes the contour
Line art carries only contour and structural lines, with no material or colour information at all. It suits two situations: keeping the skeleton of the frame without letting grayscale restrict skin tone, and work that needs clean lines — character design sheets, storyboard wireframes and the like.
Compare mode: both at once
Set “Source” to compare in the pose library‘s filter bar and hovering a card fades the line art in live; opening the full-size view puts the depth map and the line art side by side, with a small badge at the top of each side marking the source.
This step is worth taking. Some poses have a clean contour in depth and smear where the arm overlaps the torso once converted to line art; others are crisp in line art and muddy in depth. A quick look now saves a lot more than generating five wasted images and going back.
How the 288 pose sets are organised
The library holds 288 pose sets and 1,440 angles, filtered on two levels:
- System: full body single · sexy, full body single · standard, upper body single, intimate pair
- Category: standing, seated, kneeling, leaning, lying, walking and other action types
The two levels intersect. When you switch to a system, the category dropdown only keeps the options that actually have images in that system — a deliberate static build, with the filter conditions written into the page so search engines can read the full content without running scripts.
Five angles decide the camera
Each pose has five views: front, high, low, side and three-quarter. These are not five arbitrarily rotated views but five different camera languages:
- Front: the most universal, good for symmetrical actions
- High: the figure looks small and vulnerable, good for being surrounded or falling
- Low: the figure is stretched and appears taller, good for dominance and heroic stances
- Side: the clearest contour, the first choice for back actions and spine curves
- Three-quarter: between front and side, balancing volume and legibility
Do not skimp on choosing the angle. High and low produce completely different perspective, and choosing wrong makes everything after it wasted.
How to use the library
- Choose the system first and narrow the range to one group
- Filter once more by category — usually a few dozen images remain
- Lock the angle and keep only the camera position you need
- Switch to compare mode and confirm both versions of the pose hold
- Open the full-size view to download the original, or use “Bulk download” at the top right to take a whole system as a pack
Step 5 comes in two forms: a single download is the 900×1200 PNG original; a bulk download packs by “system × angle” and includes both the depth maps and the line art, 80 images per pack.
Combining it with the other libraries
Pose only answers “how to stand”. Next comes what to wear (costume library) and what the face looks like (portrait library).
You do not have to assemble them by hand — the Character Studio can pull one item from each of the three libraries into a single character recipe and generate a composite image in one click.