Edit photos by describing the change, on your own GPU. A free, open-source alternative to Nano Banana and ChatGPT image editing: photos stay on your computer, no daily limit. One installer for Windows, Mac and Linux sets up the best open model (Qwen-Image 2.1). MCP server for AI agents.
TypeScript ai-photo-editorcomfyui-alternativegenerative-fillimage-editingimage-generationinpaintinglinuxlocal-aimacosmcpmcp-servernano-banana-alternativeofflinephoto-editorprivacyqwen-imageqwen-image-2-1stable-diffusion-cppvulkanwindows
Replace the README demo with the "Fatto in casa." walkthrough
Gabriele Lozupone024cc9f28aa3
diff --git a/README.md b/README.mdindex 4733a81..1745f07 100644--- a/README.md+++ b/README.md@@ -32,7 +32,7 @@</p><p align="center">- <img src="docs/media/demo.gif" alt="Kokon: open a photo, type a sentence, get the edited photo" width="860">+ <img src="docs/media/demo.gif" alt="Kokon demo: create the Colosseum, rebuild it out of pasta, combine it into a kitchen scene, expand it and add a title" width="860"></p><p align="center">diff --git a/docs/media/demo.gif b/docs/media/demo.gifindex 7bab358..e0c71fe 100644Binary files a/docs/media/demo.gif and b/docs/media/demo.gif differdiff --git a/scripts/readme/make_demo_gif.py b/scripts/readme/make_demo_gif.pyindex 818a654..277dd5d 100644--- a/scripts/readme/make_demo_gif.py+++ b/scripts/readme/make_demo_gif.py@@ -1,20 +1,33 @@"""Builds docs/media/demo.gif from a raw screen recording of the app.-Usage: python scripts/readme/make_demo_gif.py <recording.mp4> <out.gif>+Usage: python scripts/readme/make_demo_gif.py <recording.mp4> <out.gif> [pasta|mug]Interaction plays sped up; each generation wait is squeezed to ~1.2 s with a label-showing how long it really took. Segment times are for the 2026-10-08 recording.+showing how long it really took. Segment times are per recording (found from the+brightness of the image area: waits are dimmed and flicker with the progress bar)."""import subprocess, sysSRC, OUT = sys.argv[1], sys.argv[2]-WIDTH, FPS, WAIT_S, HOLD_S = 1000, 12, 1.2, 2.5+WIDTH, FPS, WAIT_S, HOLD_S = 900, 10, 1.0, 3.0FONT = "C\\:/Windows/Fonts/segoeuib.ttf"# (start, end, speed) for interaction, or (start, end, "wait")-SEGMENTS = [- (0, 7, 2.5), (7, 27, "wait"), (27, 37, 2.5), (37, 63, "wait"), (63, 71, 2.5),- (71, 97, "wait"), (97, 102, 2.5), (102, 128, "wait"), (128, 134, 2.5),- (134, 159, "wait"), (159, 198, 4.0), (198, 238, "wait"), (238, 241.7, 1.0),-]+RECORDINGS = {+ # 2026-10-08: mug, create -> colour edits -> kitchen -> text -> expand+ "mug": [+ (0, 7, 2.5), (7, 27, "wait"), (27, 37, 2.5), (37, 63, "wait"), (63, 71, 2.5),+ (71, 97, "wait"), (97, 102, 2.5), (102, 128, "wait"), (128, 134, 2.5),+ (134, 159, "wait"), (159, 198, 4.0), (198, 238, "wait"), (238, 241.7, 1.0),+ ],+ # 2026-10-09: "Fatto in casa." Colosseum -> variation -> pasta -> sauce -> kitchen+ # -> combine -> expand -> title+ "pasta": [+ (0, 4, 3.0), (4, 23, "wait"), (23, 25, 1.0), (25, 45, "wait"), (45, 55, 3.0),+ (55, 79, "wait"), (79, 95, 2.5), (95, 120, "wait"), (120, 137, 3.5),+ (137, 157, "wait"), (157, 172, 3.0), (172, 203, "wait"), (203, 228, 4.0),+ (228, 268, "wait"), (268, 279, 3.0), (279, 320, "wait"), (320, 325.1, 2.0), # ends on the full title+ ],+}+SEGMENTS = RECORDINGS[sys.argv[3] if len(sys.argv) > 3 else "pasta"]parts, labels = [], []for i, (a, b, sp) in enumerate(SEGMENTS):@@ -30,6 +43,6 @@ for i, (a, b, sp) in enumerate(SEGMENTS):parts.append(f + f",fps={FPS}[s{i}]")labels.append(f"[s{i}]")graph = ";".join(parts) + f";{''.join(labels)}concat=n={len(labels)}:v=1:a=0,scale={WIDTH}:-1:flags=lanczos,split[x][y];" \- "[x]palettegen=max_colors=192:stats_mode=diff[p];[y][p]paletteuse=dither=bayer:bayer_scale=4:diff_mode=rectangle"+ "[x]palettegen=max_colors=192:stats_mode=diff[p];[y][p]paletteuse=dither=bayer:bayer_scale=5:diff_mode=rectangle"subprocess.run(["ffmpeg", "-v", "error", "-y", "-i", SRC, "-filter_complex", graph, "-loop", "0", OUT], check=True)print("ok", OUT)