thethreemagi/bookapp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

v1.2.0: Prefer Gemini 2.x models, improve cover generation and Docker health

Browse Source 
Model selection (ai.py):
- get_optimal_model() now scores Gemini 2.5 > 2.0 > 1.5 when ranking candidates
- get_default_models() fallbacks updated to gemini-2.0-pro-exp (logic) and gemini-2.0-flash (writer/artist)
- AI selection prompt rewritten: includes Gemini 2.x pricing context, guidance to avoid 'thinking' models for writer/artist roles, and instructions to prefer 2.x over 1.5
- Added image_model_name and image_model_source globals for UI visibility
- init_models() now reads MODEL_IMAGE_HINT; tries imagen-3.0-generate-001 then imagen-3.0-fast-generate-001 on both Gemini API and Vertex AI paths

Cover generation (marketing.py):
- Fixed display bug: "Attempt X/5" now correctly reads "Attempt X/3"
- Added imagen-3.0-fast-generate-001 as intermediate fallback before legacy Imagen 2
- Quality threshold: images with score < 5 are only kept if nothing better exists
- Smarter prompt refinement on retry: deformity, blur, and watermark critique keywords each append targeted corrections to the art prompt
- Fixed missing sys import (sys.platform check for macOS was silently broken)

Config / Docker:
- config.py: added MODEL_IMAGE_HINT env var, bumped version to 1.2.0
- docker-compose.yml: added MODEL_IMAGE environment variable
- Dockerfile: added libpng-dev and libfreetype6-dev for better font/PNG rendering; added HEALTHCHECK so Portainer detects unhealthy containers

System status UI:
- system_status.html: added Image row showing active Imagen model and provider (Gemini API / Vertex AI)
- Added cache expiry countdown with colour-coded badges

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Mike Wichers
2026-02-20 10:31:02 -05:00
parent 5e0def99c1
commit 2a9a605800
7 changed files with 171 additions and 70 deletions
Download Patch File Download Diff File Expand all files Collapse all files

112
modules/ai.py
View File

@@ -31,6 +31,8 @@ model_image = None
logic_model_name = "models/gemini-1.5-pro"
writer_model_name = "models/gemini-1.5-flash"
artist_model_name = "models/gemini-1.5-flash"
image_model_name = None
image_model_source = "None"
class ResilientModel:
def __init__(self, name, safety_settings, role):
@@ -75,10 +77,15 @@ def get_optimal_model(base_type="pro"):
candidates = [m.name for m in models if base_type in m.name]
if not candidates: return f"models/gemini-1.5-{base_type}"
def score(n):
# Prioritize stable models (higher quotas) over experimental/beta ones
if "exp" in n or "beta" in n or "preview" in n: return 0
if "latest" in n: return 50
return 100
# Prefer newer generations: 2.5 > 2.0 > 1.5
gen_bonus = 0
if "2.5" in n: gen_bonus = 300
elif "2.0" in n: gen_bonus = 200
elif "2." in n: gen_bonus = 150
# Within a generation, prefer stable over experimental
if "exp" in n or "beta" in n or "preview" in n: return gen_bonus + 0
if "latest" in n: return gen_bonus + 50
return gen_bonus + 100
return sorted(candidates, key=score, reverse=True)[0]
except Exception as e:
utils.log("SYSTEM", f"⚠️ Error finding optimal model: {e}")
@@ -86,9 +93,9 @@ def get_optimal_model(base_type="pro"):
def get_default_models():
return {
"logic": {"model": "models/gemini-1.5-pro", "reason": "Fallback: Default Pro model selected.", "estimated_cost": "$3.50/1M"},
"writer": {"model": "models/gemini-1.5-flash", "reason": "Fallback: Default Flash model selected.", "estimated_cost": "$0.075/1M"},
"artist": {"model": "models/gemini-1.5-flash", "reason": "Fallback: Default Flash model selected.", "estimated_cost": "$0.075/1M"},
"logic": {"model": "models/gemini-2.0-pro-exp", "reason": "Fallback: Gemini 2.0 Pro for complex reasoning and JSON adherence.", "estimated_cost": "$0.00/1M (Experimental)"},
"writer": {"model": "models/gemini-2.0-flash", "reason": "Fallback: Gemini 2.0 Flash for fast, high-quality creative writing.", "estimated_cost": "$0.10/1M"},
"artist": {"model": "models/gemini-2.0-flash", "reason": "Fallback: Gemini 2.0 Flash for visual prompt design.", "estimated_cost": "$0.10/1M"},
"ranking": []
}
@@ -131,29 +138,37 @@ def select_best_models(force_refresh=False):
model = genai.GenerativeModel(bootstrapper)
prompt = f"""
ROLE: AI Model Architect
TASK: Select the optimal Gemini models for specific application roles.
TASK: Select the optimal Gemini models for a book-writing application. Prefer newer Gemini 2.x models when available.
AVAILABLE_MODELS:
{json.dumps(models)}
PRICING_CONTEXT (USD per 1M tokens):
- Flash Models (e.g. gemini-1.5-flash): ~$0.075 Input / $0.30 Output. (Very Cheap)
- Pro Models (e.g. gemini-1.5-pro): ~$3.50 Input / $10.50 Output. (Expensive)
PRICING_CONTEXT (USD per 1M tokens, approximate):
- Gemini 2.5 Pro/Flash: Best quality/speed; check current pricing.
- Gemini 2.0 Flash: ~$0.10 Input / $0.40 Output. (Fast, cost-effective, excellent quality).
- Gemini 2.0 Pro Exp: Free experimental tier with strong reasoning.
- Gemini 1.5 Flash: ~$0.075 Input / $0.30 Output. (Legacy, still reliable).
- Gemini 1.5 Pro: ~$1.25 Input / $5.00 Output. (Legacy, expensive).
CRITERIA:
- LOGIC: Needs complex reasoning, JSON adherence, and instruction following. (Prefer Pro/1.5).
- WRITER: Needs creativity, prose quality, and speed. (Prefer Flash/1.5 for speed, or Pro for quality).
- ARTIST: Needs visual prompt understanding.
- LOGIC: Needs complex reasoning, strict JSON adherence, plot consistency, and instruction following.
-> Prefer: Gemini 2.5 Pro > 2.0 Pro > 2.0 Flash > 1.5 Pro
- WRITER: Needs creativity, prose quality, long-form text generation, and speed.
-> Prefer: Gemini 2.5 Flash/Pro > 2.0 Flash > 1.5 Flash (balance quality/cost)
- ARTIST: Needs rich visual description, prompt understanding for cover art design.
-> Prefer: Gemini 2.0 Flash > 1.5 Flash (speed and visual understanding)
CONSTRAINTS:
- Avoid 'experimental' or 'preview' unless no stable version exists.
- Prioritize 'latest' or stable versions.
OUTPUT_FORMAT (JSON):
- Strongly prefer Gemini 2.x over 1.5 where available.
- Avoid 'experimental' or 'preview' only if a stable 2.x version exists; otherwise experimental 2.x is fine.
- 'thinking' models are too slow/expensive for Writer/Artist roles.
- Provide a ranking of ALL available models from best to worst overall.
OUTPUT_FORMAT (JSON only, no markdown):
{{
"logic": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX Input / $X.XX Output" }},
"writer": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX Input / $X.XX Output" }},
"artist": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX Input / $X.XX Output" }},
"logic": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX/1M" }},
"writer": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX/1M" }},
"artist": {{ "model": "string", "reason": "string", "estimated_cost": "$X.XX/1M" }},
"ranking": [ {{ "model": "string", "reason": "string", "estimated_cost": "string" }} ]
}}
"""
@@ -195,7 +210,7 @@ def select_best_models(force_refresh=False):
return fallback
def init_models(force=False):
global model_logic, model_writer, model_artist, model_image, logic_model_name, writer_model_name, artist_model_name
global model_logic, model_writer, model_artist, model_image, logic_model_name, writer_model_name, artist_model_name, image_model_name, image_model_source
if model_logic and not force: return
genai.configure(api_key=config.API_KEY)
@@ -264,13 +279,28 @@ def init_models(force=False):
model_writer.update(writer_name)
model_artist.update(artist_name)
# Initialize Image Model (Default to None)
# Initialize Image Model
model_image = None
image_model_name = None
image_model_source = "None"
hint = config.MODEL_IMAGE_HINT if hasattr(config, 'MODEL_IMAGE_HINT') else "AUTO"
if hasattr(genai, 'ImageGenerationModel'):
try: model_image = genai.ImageGenerationModel("imagen-3.0-generate-001")
except: pass
img_source = "Gemini API" if model_image else "None"
# Candidate image models in preference order
if hint and hint != "AUTO":
candidates = [hint]
else:
candidates = ["imagen-3.0-generate-001", "imagen-3.0-fast-generate-001"]
for candidate in candidates:
try:
model_image = genai.ImageGenerationModel(candidate)
image_model_name = candidate
image_model_source = "Gemini API"
utils.log("SYSTEM", f"✅ Image model: {candidate} (Gemini API)")
break
except Exception:
continue
# Auto-detect GCP Project from credentials if not set (Fix for Image Model)
if HAS_VERTEX and not config.GCP_PROJECT and config.GOOGLE_CREDS and os.path.exists(config.GOOGLE_CREDS):
@@ -326,9 +356,17 @@ def init_models(force=False):
utils.log("SYSTEM", f"✅ Vertex AI initialized (Project: {config.GCP_PROJECT})")
# Override with Vertex Image Model if available
try:
model_image = VertexImageModel.from_pretrained("imagen-3.0-generate-001")
img_source = "Vertex AI"
except: pass
utils.log("SYSTEM", f"Image Generation Provider: {img_source}")
vertex_candidates = ["imagen-3.0-generate-001", "imagen-3.0-fast-generate-001"]
if hint and hint != "AUTO":
vertex_candidates = [hint]
for candidate in vertex_candidates:
try:
model_image = VertexImageModel.from_pretrained(candidate)
image_model_name = candidate
image_model_source = "Vertex AI"
utils.log("SYSTEM", f"✅ Image model: {candidate} (Vertex AI)")
break
except Exception:
continue
utils.log("SYSTEM", f"Image Generation Provider: {image_model_source} ({image_model_name or 'unavailable'})")

71
modules/marketing.py
View File

@@ -1,10 +1,10 @@
import os
import sys
import json
import shutil
import textwrap
import subprocess
import requests
import google.generativeai as genai
from . import utils
import config
from modules import ai
@@ -212,59 +212,82 @@ def generate_cover(bp, folder, tracking=None, feedback=None, interactive=False):
best_img_score = 0
best_img_path = None
MAX_IMG_ATTEMPTS = 3
if regenerate_image:
for i in range(1, 4):
utils.log("MARKETING", f"Generating cover art (Attempt {i}/5)...")
for i in range(1, MAX_IMG_ATTEMPTS + 1):
utils.log("MARKETING", f"Generating cover art (Attempt {i}/{MAX_IMG_ATTEMPTS})...")
try:
if not ai.model_image: raise ImportError("No Image Generation Model available.")
status = "success"
try:
result = ai.model_image.generate_images(prompt=art_prompt, number_of_images=1, aspect_ratio=ar)
except Exception as e:
if "resource" in str(e).lower() and ai.HAS_VERTEX:
utils.log("MARKETING", "⚠️ Imagen 3 failed. Trying Imagen 2...")
fb_model = ai.VertexImageModel.from_pretrained("imagegeneration@006")
result = fb_model.generate_images(prompt=art_prompt, number_of_images=1, aspect_ratio=ar)
status = "success_fallback"
else: raise e
err_lower = str(e).lower()
# Try fast imagen variant before falling back to legacy
if ai.HAS_VERTEX and ("resource" in err_lower or "quota" in err_lower):
try:
utils.log("MARKETING", "⚠️ Imagen 3 failed. Trying Imagen 3 Fast...")
fb_model = ai.VertexImageModel.from_pretrained("imagen-3.0-fast-generate-001")
result = fb_model.generate_images(prompt=art_prompt, number_of_images=1, aspect_ratio=ar)
status = "success_fast"
except Exception:
utils.log("MARKETING", "⚠️ Imagen 3 Fast failed. Trying Imagen 2...")
fb_model = ai.VertexImageModel.from_pretrained("imagegeneration@006")
result = fb_model.generate_images(prompt=art_prompt, number_of_images=1, aspect_ratio=ar)
status = "success_fallback"
else:
raise e
attempt_path = os.path.join(folder, f"cover_art_attempt_{i}.png")
result.images[0].save(attempt_path)
utils.log_usage(folder, "imagen", image_count=1)
score, critique = evaluate_image_quality(attempt_path, art_prompt, ai.model_writer, folder)
if score is None: score = 0
utils.log("MARKETING", f" -> Image Score: {score}/10. Critique: {critique}")
utils.log_image_attempt(folder, "cover", art_prompt, f"cover_art_{i}.png", status, score=score, critique=critique)
if interactive:
# Open image for review
try:
if os.name == 'nt': os.startfile(attempt_path)
elif sys.platform == 'darwin': subprocess.call(('open', attempt_path))
else: subprocess.call(('xdg-open', attempt_path))
except: pass
if Confirm.ask(f"Accept cover attempt {i} (Score: {score})?", default=True):
best_img_path = attempt_path
break
else:
utils.log("MARKETING", "User rejected cover. Retrying...")
continue
if score > best_img_score:
# Only keep as best if score meets minimum quality bar
if score >= 5 and score > best_img_score:
best_img_score = score
best_img_path = attempt_path
if score == 10:
utils.log("MARKETING", " -> Perfect image accepted.")
elif best_img_path is None and score > 0:
# Accept even low-quality image if we have nothing else
best_img_score = score
best_img_path = attempt_path
if score >= 9:
utils.log("MARKETING", " -> High quality image accepted.")
break
if "scar" in critique.lower() or "deform" in critique.lower() or "blur" in critique.lower():
art_prompt += " (Ensure high quality, clear skin, no scars, sharp focus)."
# Refine prompt based on critique keywords
prompt_additions = []
critique_lower = critique.lower() if critique else ""
if "scar" in critique_lower or "deform" in critique_lower:
prompt_additions.append("perfect anatomy, no deformities")
if "blur" in critique_lower or "blurry" in critique_lower:
prompt_additions.append("sharp focus, highly detailed")
if "text" in critique_lower or "letter" in critique_lower:
prompt_additions.append("no text, no letters, no watermarks")
if prompt_additions:
art_prompt += f". ({', '.join(prompt_additions)})"
except Exception as e:
utils.log("MARKETING", f"Image generation failed: {e}")
if "quota" in str(e).lower(): break

3
modules/web_app.py
View File

@@ -1303,7 +1303,8 @@ def system_status():
models_info = cache_data.get('models', {})
except: pass
return render_template('system_status.html', models=models_info, cache=cache_data, datetime=datetime)
return render_template('system_status.html', models=models_info, cache=cache_data, datetime=datetime,
image_model=ai.image_model_name, image_source=ai.image_model_source)
@app.route('/personas')
@login_required