Natal Chart Image

Most nodes in the Comfyui-Kerykeion pack spit out JSON. This one spits out a picture: the classic zodiac wheel, computed from a real natal chart, as an IMAGE tensor you can preview, save, or feed straight into ControlNet. The pack's README's pitch is the fun part - connect it to a Canny or Lineart ControlNet and let Stable Diffusion turn the actual geometry of your birth chart into mystical art.

That's a surprisingly neat trick for ComfyUI: normally the only way to get a chart wheel is a website or an astrology app. Here it's a first-class node in your graph.

How it works

Same three-step pipeline as Natal Chart Calculator - geocode the city with Nominatim, resolve the timezone with timezonefinder, then build a kerykeion AstrologicalSubject in offline mode (local Swiss Ephemeris math). The difference is what happens after:

1. SVG generation. Kerykeion draws the chart as a vector graphic (KerykeionChartSVG), producing an SVG string.
2. SVG → PNG. cairosvg rasterizes that SVG into pixel data.
3. PNG → tensor. PIL loads the image and it becomes a float tensor, scaled to 0–1, shaped like any ComfyUI image you've seen.

That SVG-middle-step is why this node has a dependency the others don't: cairosvg, and through it the native Cairo graphics library on Linux.

The inputs that matter

Identical to Natal Chart Calculator - year, month, day, hour, minute, city_name - and the same caveats apply: the birth time matters for the Ascendant and house cusps, and the city is resolved by a live geocoder, so it needs internet and a findable city.

The single output

• natal_chart_image - an IMAGE tensor. The three things people do with it:

1. Preview it. Wire it into a Preview Image or Save Image node just to look at the wheel.
2. ControlNet it. This is the fun one. Feed the chart image through a Canny or Lineart preprocessor into a ControlNet Apply node, then prompt SD with something like "a beautiful mystical tarot card, golden lines, deep cosmic background". The wheel's real geometry constrains the composition. (If you're on a union-style model, mind the recommended control weight - modern unions overcook at the old 1.0 default.)
3. Composite it. Save it and overlay it on artwork downstream as a genuinely accurate chart graphic.

Installing it

Same pack install as everything else:

cd ComfyUI/custom_nodes
git clone https://github.com/Meisoftcoltd/Comfyui-Kerykeion
cd Comfyui-Kerykeion
pip install -r requirements.txt

or "ComfyUI-Kerykeion" via ComfyUI Manager, then restart. No model files to download. The one extra gotcha is Cairo itself: on Linux, if cairosvg fails to import, you usually need the system library first (sudo apt install libcairo2 on Debian/Ubuntu). On Windows the wheels usually bundle it. If this node errors while the JSON nodes work, cairosvg is the first suspect.

Common issues

• You get a black image and no error. This is the trap, and it's specific to this node. Where the calculators return a JSON error string, Natal Chart Image returns a blank 512×512 black tensor on any failure - city not found, timezone not found, any exception. No red bubble, no console message you'd definitely notice. If you render a wheel and get black, check the console for Error (NatalChartImageNode) and verify the city actually geocoded. This is also worth knowing if you pipe the output to ControlNet: a black control image produces garbage conditionally-guided generations, so preview the wheel before you trust it.
• Geocoding rate limits. Same as the rest of the pack - every run hits OpenStreetMap's free Nominatim, which is polite about being hammered but not infinite.
• Fixed resolution. There's no width/height control; you get whatever kerykeion's SVG template renders at. If you need a bigger wheel for ControlNet, upscale it downstream rather than expecting a resolution setting here.

It's the pack's showpiece node, and honestly the most distinctive thing in it - an actual, correct astrological chart wheel as a composable image. Just remember the black-frame failure mode and preview before you ControlNet.