Directional Tiles

Geometry-driven directional prompts using polyhedra faces and Qwen2.5-VL vision

Last updated: 2026-10-27

Alpha — available for testing and feedback. Controls and behavior may change.

What It Is

Directional prompts assign text to aimed points around the sphere. You can write them by hand without vision analysis. Optional Describe uses the local Qwen vision model to suggest text from the selected image/crop.

Each polyhedron face points to a specific direction on the sphere. You assign a prompt per face, and the AI generates content matching each direction -- different sky, ground, and everything between.

How It Works

The equirectangular image is divided into directional zones based on polyhedra geometry:

  1. Pick a polyhedron -- 14 shapes available, from Cube (6 faces) to Goldberg GP(3,0) (92 faces). More sample directions mean more descriptions to inspect; they do not guarantee higher output quality.
  2. Assign prompts per face -- each face maps to a {yaw, pitch, prompt} point. Editable, deletable, freely reorderable.
  3. Prompts mix across the sphere -- between face centers, normalized Gaussian influence combines nearby prompt anchors.

Available shapes and their face counts:

Shape Faces Best For
Cube 6 Quick cardinal sweeps
Dodecahedron 12 Fast expressive coverage
Rhombic Dodecahedron 12 Alternate 12-direction layout
Truncated Octahedron 14 Balanced directional spread
Icosahedron 20 Standard polyhedra coverage
Rhombicuboctahedron 26 Dense sampling
Truncated Icosahedron 32 More sample directions
Chamfered Dodecahedron 42 Very dense coverage
Disdyakis Dodecahedron 48 Ultra-dense hemisphere
Pentakis Dodecahedron 60 Fine-grain directional
Deltoidal Hexecontahedron 60 Alternate 60-face layout
Rhombicosidodecahedron 62 Maximum traditional polyhedra
Goldberg GP(2,1) 72 Dense sampling
Goldberg GP(3,0) 92 Dense sampling

PolyhedraTileControls

The Tiles list shows every face-direction tile. Each card displays its direction (yaw/pitch), assigned prompt, and Describe status. The same list is used for Prompt tiles and for Spot rows.

  • Select -- click a tile's metadata bar or Target icon to load it into the editor. A single click does not move the 360 camera.
  • Look at -- double-click the metadata bar or Target icon to point the 360 viewer at that tile. The view glides using your Camera smoothing setting. Zoom (FOV) does not change. Face-only tiles with no yaw/pitch do nothing on double-click. In the compact list, Enter on a selected posed row does the same.
  • Aim overlay -- on the 360 HUD, the selected marker shows its current mix share. Show all adds every other marker whose share survives Cutoff. Right-click blank space to Add Spot Here; right-click a marker to Duplicate Spot or Delete Spot.
  • Polyhedron preview -- a rotatable solid view showing all face center points. Hover or click faces to see their prompts.
  • Editable point list -- every generated point is editable and deletable. Add or remove directions freely.

Qwen2.5-VL Describe

Click Describe on a tile to have Qwen2.5-VL analyze what is currently visible in that direction and suggest a prompt. No API key required -- everything runs locally on your GPU.

  • How it works -- the app perspective-crops your equirectangular input at the face-center direction, runs Qwen2.5-VL on the crop, and writes a prompt describing the content.
  • Batch describe -- use Describe All to run Qwen2.5-VL on every face in one batched session. All directional prompts populate at once.
  • Install the vision model through the app's requirements/model controls and wait for GPU readiness. Describe is local and keyless, but still needs its weights and hardware resources.

Gaussian prompt interpolation

Each render tile gets a normalized Gaussian mix based on angular distance and neighbouring anchor spacing. Reach sets the influence width; Auto Reach derives it from nearby interpolation anchors. Cutoff removes low-share prompts from the mix. This is not cosine weighting.

Semantic Blend, when available in the selected pipeline, changes how prompt influences combine. Compare it on a small test; neither blend path guarantees consistent scene structure.

Image Offset Interaction

Directional prompts stay anchored to absolute sphere directions. When you offset (pan/rotate) the image, the image moves under the fixed prompt grid:

  • Your "sky" prompt always points up, regardless of where the image content sits.
  • Your "ground" prompt always points down.
  • Merely looking around does not move the authored directions.
  • Rotating the image underneath the grid can make a previously described prompt point at different content. Inspect alignment and move/edit or re-describe the relevant points when that happens.

Image rotation, prompt placement and tile processing order are different choices; inspect each deliberately. See Image Offset.

Generate Flow

  1. Pick a polyhedron -- select the shape that matches your desired directional granularity.
  2. Place directional prompts -- type prompts per face, or leave some blank to inherit the global prompt.
  3. Optional: Describe -- click Describe or Describe All to let Qwen2.5-VL analyze each direction and suggest prompts.
  4. Render -- use RENDER for the configured pass, or the selected tile's supported preview/generation action. Prompt text generation and rendering pixels are different actions.
  5. Review -- check the result in the viewer. Edit prompts and regenerate specific directions as needed.

Tile Ordering Strategies

15 strategies control the order in which tiles are generated. The order affects how adjacent tiles influence each other during processing:

Strategy Behavior
original Template's default order
latitude_rings Top to bottom by latitude
latitude_rings_reverse Bottom to top by latitude
spiral Spiral outward from center
radial_from_point Radiating from a chosen point
custom Manually specified order
sweep_front_back Front to back sweep
sweep_back_front Back to front sweep
sweep_left_right Left to right sweep
sweep_right_left Right to left sweep
equator_out Equator outward to poles
poles_in Poles inward to equator
max_spread Maximize distance between consecutive tiles
checkerboard Alternating checkerboard pattern
random Randomized order

Works with all visible diffusion pipelines (SDXL, SD3.5, SD2.1, SD1.5, FLUX, Z-Image Turbo, and Qwen-Image in Alpha builds).

When placing or inspecting directional points in the viewer/navigator:

  • Look at a tile from the Tiles list (double-click, or compact Enter) to glide the 360 view to that direction without changing FOV.
  • FOV changes the perspective crop used by Describe. It does not change prompt mixing.
  • Reach controls Gaussian influence width. Cutoff removes small shares. Use those controls when a direction should affect more or less of the sphere.
  • Place-and-describe workflows: drop a direction, write its prompt, then let auto-fill or AI transform expand the set — product language is Directional.

Upscale vocabulary (when chaining after directional work)

Upscaling has several workflow surfaces. On an open render, one Resize toolbar popover now contains both absolute Size destinations and native-relative Upscale factors; there is no separate Upscale toolbar button.

Path What it is
Viewer Resize → Upscale Post-render Real-ESRGAN Alpha on the viewed graph node; 2× / 4× / 8× / 16× targets up to the edge cap
Upscale module Real-ESRGAN Alpha in the render module chain — any valid license
Real-ESRGAN pipeline Alpha standalone processor, not the module defaults
Graph lineage resize nodes for absolute Size actions and upscale nodes for AI scale actions
360-LoRA cell upscale Prep-time upscale on LoRA input, separate from final module upscale

See Upscale.

Next Steps

Esc
Searching...
No results found.
Type to search the documentation
↑↓ Navigate ↵ Select Esc Close

We use cookies for analytics to understand how visitors use our site. Privacy Policy