From 37bf5281c3992f6047c9d61c1612e8af004532c4 Mon Sep 17 00:00:00 2001 From: mike Date: Fri, 26 Jun 2026 03:26:27 +0200 Subject: [PATCH] outpaint orbit --- backlog.md | 1 + tour-comfy/car.html | 35 ++++++++++++++++++------------ tour-comfy/edit_api.py | 48 ++++++++++++------------------------------ tour-comfy/poses.md | 1 + 4 files changed, 38 insertions(+), 47 deletions(-) diff --git a/backlog.md b/backlog.md index b873ad1..10ff247 100644 --- a/backlog.md +++ b/backlog.md @@ -88,5 +88,6 @@ in car.html - studio view we have a action to pad the image we also have this fe 2) we have a bug that when pressing the ui-action "duplicate" and "pad" the browser needs to refresh before we see the result. Other actions like generating from pose do not hae this bug, fix the bug and boy scout adjecnt actions if they are missing the required technical follow up. Fix the bug for the features. 3) In the "Generate" tab, the CUSTOM prompt is shown twice, above POSES and statically always in view in the right bottom. Remove the statically CUSTOM prompt from the UI. 4) The Generate button is always visible in the "Generate" tab, however we cannot jump to it via keyboard. Add a keyboard shortcut to jump to the Generate button "G" or simialler. +5) remove the "set prompt" Improve the UI/UX for the end-user experience. Where effort is low and impact is high. Use own insight for further refinement of these improvements. diff --git a/tour-comfy/car.html b/tour-comfy/car.html index 5bf3df3..6c83ff4 100644 --- a/tour-comfy/car.html +++ b/tour-comfy/car.html @@ -3413,7 +3413,7 @@ if (job.status === 'done') { showToast('Pose generation done!', 'success'); _followLatestGid = lbCurrentGid; - refreshNow(); + refreshNow(true); } else { showToast('Pose generation failed: ' + (job.error || 'unknown'), 'error'); } @@ -3885,7 +3885,7 @@ updateStudio(); _followLatestGid = lbCurrentGid; - setTimeout(refreshNow, 500); // give server time to write static JSON + refreshNow(true); } else { showToast('Padded', 'success'); lbUrls[lbIdx] = IMAGE_FOLDER + fname + '?t=' + Date.now(); @@ -3936,7 +3936,7 @@ updateStudio(); _followLatestGid = lbCurrentGid; - setTimeout(refreshNow, 500); + refreshNow(true); } else { showToast('Duplicate failed', 'error'); } @@ -3955,7 +3955,7 @@ showToast('Background removed', 'success'); fileHasBg[fname] = false; _bustStudioImage(fname); // cache-busts viewer + calls updateStudio() - refreshNow(); + refreshNow(true); } else { showToast(`Failed to remove background for ${fname}`, 'error'); } @@ -3977,7 +3977,7 @@ showToast('Transparency inverted', 'success'); fileHasBg[fname] = false; _bustStudioImage(fname); - refreshNow(); + refreshNow(true); } else { showToast('Invert failed: ' + await r.text(), 'error'); } @@ -4006,7 +4006,7 @@ const d = await r.json(); showToast(`Video created: ${d.output}`, 'success'); _followLatestGid = lbCurrentGid; - refreshNow(); + refreshNow(true); } catch (e) { showToast('Video error: ' + e, 'error'); } if (btn) { btn.disabled = false; btn.textContent = 'Make video'; } } @@ -4659,7 +4659,7 @@ }); } - async function loadImages() { + async function loadImages(bypassStatic = false) { if (loadImages._inFlight) return; loadImages._inFlight = true; const gallery = document.getElementById('gallery'); @@ -4671,6 +4671,7 @@ // Primary: pre-generated static JSON (fast, no DB round-trip per request) try { + if (bypassStatic) throw new Error("Bypass static requested"); const r = await fetch(`${API}/output/_data/images.json?t=${Date.now()}`, { signal: AbortSignal.timeout(4000) }); if (r.ok) { const data = await r.json(); @@ -4914,9 +4915,14 @@ // Direct callers that need a response (pollJob completion, etc.) still use // loadImages() directly; everything else should go through refreshNow(). let _refreshTimer = null; - function refreshNow() { + let _refreshBypassStatic = false; + function refreshNow(bypass = false) { + if (bypass) _refreshBypassStatic = true; clearTimeout(_refreshTimer); - _refreshTimer = setTimeout(loadImages, 80); + _refreshTimer = setTimeout(() => { + loadImages(_refreshBypassStatic || !!_followLatestGid); + _refreshBypassStatic = false; + }, 80); } const API = 'http://127.0.0.1:8500'; @@ -6010,7 +6016,7 @@ const poll = () => setTimeout(async () => { try { const s = await fetch(`${API}/batch/${job_id}`).then(r => r.json()); - if (s.status === 'done') { refreshNow(); } + if (s.status === 'done') { refreshNow(true); } else if (s.status !== 'error' && s.status !== 'cancelled') poll(); } catch(e) {} }, 2000); @@ -6168,7 +6174,7 @@ if (job.done > prevDone) { prevDone = job.done; _followLatestGid = lbCurrentGid; - refreshNow(); + refreshNow(true); } if (job.status === 'running') { pollGen(); return; } if (job.status === 'cancelled') { @@ -6188,7 +6194,7 @@ _sbPoseEdits = {}; _sbGenJobId = null; _followLatestGid = lbCurrentGid; - refreshNow(); + refreshNow(true); renderSidebarGenerate(); } else { textEl.textContent = `Error: ${job.error || 'unknown'}`; @@ -6445,7 +6451,7 @@ if (btn) btn.textContent = 'Done ✓'; showToast(`Faceswap complete: ${job.output || ''}`, 'success'); _followLatestGid = lbCurrentGid; - setTimeout(() => { refreshNow(); }, 1800); + refreshNow(true); } else { if (textEl) textEl.textContent = `Error: ${job.error || 'unknown'}`; if (btn) { btn.disabled = false; btn.textContent = 'Retry'; } @@ -7068,6 +7074,7 @@ const label = d.used_sam2 ? 'Done (SAM2)' : 'Done (rembg)'; if (statusEl) statusEl.textContent = label; showToast('Background removed', 'success'); + refreshNow(true); // Build URL same way as all gallery images: IMAGE_FOLDER + filename const nobgUrl = IMAGE_FOLDER + d.nobg_filename + '?t=' + Date.now(); _nobgSidecars.set(fname, nobgUrl); @@ -7116,6 +7123,7 @@ const d = await r.json(); if (statusEl) statusEl.textContent = 'Done (rembg)'; showToast('Background removed (rembg)', 'success'); + refreshNow(true); const nobgUrl = IMAGE_FOLDER + d.nobg_filename + '?t=' + Date.now(); _nobgSidecars.set(fname, nobgUrl); const preload = new Image(); @@ -7152,6 +7160,7 @@ lbNames = [fname, ...otherNames]; lbIdx = 0; updateStudio(); + refreshNow(true); // Fire-and-forget face extraction for face-book fetch(`${API}/images/${encodeURIComponent(fname)}/extract-face`, { method: 'POST' }) .then(r => r.ok ? r.json() : null) diff --git a/tour-comfy/edit_api.py b/tour-comfy/edit_api.py index b397b13..b446ead 100644 --- a/tour-comfy/edit_api.py +++ b/tour-comfy/edit_api.py @@ -335,36 +335,14 @@ def _target_size(w: int, h: int, max_area: int) -> tuple[int, int]: def _prep_image(pil: Image.Image, max_area: int) -> tuple[Image.Image, int, int]: """ Prepare image for ComfyUI: - 1. If area > max_area, crop from bottom if height remains >= 256. - 2. Otherwise scale (up or down) to fit area while preserving aspect. - 3. Ensure dimensions are rounded to 16. + 1. Scale (up or down) to fit area while preserving aspect. + 2. Ensure dimensions are rounded to 16. """ w, h = pil.width, pil.height - if w * h > max_area: - # Try to keep width and crop height from bottom - rw = _round16(w) - th = max_area // rw - if th >= 256: - rh = (th // 16) * 16 - if rh < 16: rh = 16 - - # To avoid black bars from .crop((0,0,rw,rh)) when rw > w, - # we crop to original w first, then resize to rw. - pil = pil.crop((0, 0, w, min(h, (rh * w) // rw))) - pil = pil.resize((rw, rh), resample=Image.LANCZOS) - return pil, rw, rh - else: - # Too wide to keep width and have decent height, scale both down - rw, rh = _target_size(w, h, max_area) - pil = pil.resize((rw, rh), resample=Image.LANCZOS) - return pil, rw, rh - else: - # Fits or is too small: scale UP to match the max_area budget - # (Legacy behavior that gives better model performance) - rw, rh = _target_size(w, h, max_area) - if rw != w or rh != h: - pil = pil.resize((rw, rh), resample=Image.LANCZOS) - return pil, rw, rh + rw, rh = _target_size(w, h, max_area) + if rw != w or rh != h: + pil = pil.resize((rw, rh), resample=Image.LANCZOS) + return pil, rw, rh def _comfy_upload(img_bytes: bytes, filename: str) -> str: @@ -1342,8 +1320,8 @@ def _invalidate_static() -> None: def _batch_worker(job_id: str, filenames: list, prompts: list[str], poses: list, seed: int, max_area: int, group_id: str | None = None, wireframe_ref: str | None = None, wireframe_time: float = 0.5, - pad_top: int = 0, pad_right: int = 0, - pad_bottom: int = 0, pad_left: int = 0, pad_fill: str = "black", + pad_top: Any = 0, pad_right: Any = 0, + pad_bottom: Any = 0, pad_left: Any = 0, pad_fill: str = "black", pad_outpaint: bool = False): output_dir = _load_output_dir() for fname in filenames: @@ -1384,6 +1362,8 @@ def _batch_worker(job_id: str, filenames: list, prompts: list[str], poses: list, cap.release() if ret: pose_guide_pil = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)) + if any([pad_top, pad_right, pad_bottom, pad_left]): + pose_guide_pil = _apply_manual_pad(pose_guide_pil, pad_top, pad_right, pad_bottom, pad_left, pad_fill) print(f"[batch] using wireframe {wireframe_ref} frame {target_frame}/{total_frames}") except Exception as wf_err: print(f"[batch] wireframe extract error: {wf_err}") @@ -1451,8 +1431,8 @@ def _batch_worker(job_id: str, filenames: list, prompts: list[str], poses: list, def _multi_ref_worker(job_id: str, filenames: list[str], prompts: list[str], poses: list, seed: int, max_area: int, - pad_top: int = 0, pad_right: int = 0, - pad_bottom: int = 0, pad_left: int = 0, + pad_top: Any = 0, pad_right: Any = 0, + pad_bottom: Any = 0, pad_left: Any = 0, pad_fill: str = "black", pad_outpaint: bool = False): """Generate one output image per prompt using filenames[0] as primary and the rest as extra refs.""" output_dir = _load_output_dir() @@ -1467,9 +1447,9 @@ def _multi_ref_worker(job_id: str, filenames: list[str], prompts: list[str], pos jobs[job_id]["status"] = "done" return - # Apply padding to the primary image if requested + # Apply padding to images if requested if any([pad_top, pad_right, pad_bottom, pad_left]): - pils[0] = (pils[0][0], _apply_manual_pad(pils[0][1], pad_top, pad_right, pad_bottom, pad_left, pad_fill)) + pils = [(name, _apply_manual_pad(pil, pad_top, pad_right, pad_bottom, pad_left, pad_fill)) for name, pil in pils] # Output group: reuse shared group if all sources belong to the same one, else new group source_groups = set() diff --git a/tour-comfy/poses.md b/tour-comfy/poses.md index cb93f39..6414919 100644 --- a/tour-comfy/poses.md +++ b/tour-comfy/poses.md @@ -4479,3 +4479,4 @@ Anatomically precise, hyperrealistic, high detail, keep the characteristics of t # Upper Body Restraining Restrain more, push person into position on her toes, and chest forward out without restraining the chest, but via the arms and elbows. Use smart steel and restraining connections but enough to immobilize for maximum exposure. +