image gen mess
This commit is contained in:
@@ -218,17 +218,41 @@ def process_image_generation_queue(self, image_ids: list, account_id: int = None
|
||||
image_type = config.get('image_type') or global_settings.image_style
|
||||
image_format = config.get('image_format', 'webp')
|
||||
|
||||
# Style to prompt enhancement mapping
|
||||
# These style descriptors are added to the image prompt for better results
|
||||
STYLE_PROMPT_MAP = {
|
||||
# Runware styles
|
||||
'photorealistic': 'ultra realistic photography, natural lighting, real world look, photorealistic',
|
||||
'illustration': 'digital illustration, clean lines, artistic style, modern illustration',
|
||||
'3d_render': 'computer generated 3D render, modern polished 3D style, depth and dramatic lighting',
|
||||
'minimal_flat': 'minimal flat design, simple shapes, flat colors, modern graphic design aesthetic',
|
||||
'artistic': 'artistic painterly style, expressive brushstrokes, hand painted aesthetic',
|
||||
'cartoon': 'cartoon stylized illustration, playful exaggerated forms, animated character style',
|
||||
# DALL-E styles (mapped from OpenAI API style parameter)
|
||||
'natural': 'natural realistic style',
|
||||
'vivid': 'vivid dramatic hyper-realistic style',
|
||||
# Legacy fallbacks
|
||||
'realistic': 'ultra realistic photography, natural lighting, photorealistic',
|
||||
}
|
||||
|
||||
# Get the style description for prompt enhancement
|
||||
style_description = STYLE_PROMPT_MAP.get(image_type, STYLE_PROMPT_MAP.get('photorealistic'))
|
||||
logger.info(f"[process_image_generation_queue] Style: {image_type} -> prompt enhancement: {style_description[:50]}...")
|
||||
|
||||
# Model-specific landscape sizes (square is always 1024x1024)
|
||||
# Based on Runware documentation for optimal results per model
|
||||
# For Runware models - based on Runware documentation for optimal results per model
|
||||
# For OpenAI DALL-E 3 - uses 1792x1024 for landscape
|
||||
MODEL_LANDSCAPE_SIZES = {
|
||||
'runware:97@1': '1280x768', # Hi Dream Full landscape
|
||||
'bria:10@1': '1344x768', # Bria 3.2 landscape (16:9)
|
||||
'google:4@2': '1376x768', # Nano Banana landscape (16:9)
|
||||
'dall-e-3': '1792x1024', # DALL-E 3 landscape
|
||||
'dall-e-2': '1024x1024', # DALL-E 2 only supports square
|
||||
}
|
||||
DEFAULT_SQUARE_SIZE = '1024x1024'
|
||||
|
||||
# Get model-specific landscape size for featured images
|
||||
model_landscape_size = MODEL_LANDSCAPE_SIZES.get(model, '1280x768')
|
||||
model_landscape_size = MODEL_LANDSCAPE_SIZES.get(model, '1792x1024' if provider == 'openai' else '1280x768')
|
||||
|
||||
# Featured image always uses model-specific landscape size
|
||||
featured_image_size = model_landscape_size
|
||||
@@ -398,7 +422,7 @@ def process_image_generation_queue(self, image_ids: list, account_id: int = None
|
||||
# Calculate actual template length with placeholders filled
|
||||
# Format template with dummy values to measure actual length
|
||||
template_with_dummies = image_prompt_template.format(
|
||||
image_type=image_type,
|
||||
image_type=style_description, # Use actual style description length
|
||||
post_title='X' * len(post_title), # Use same length as actual post_title
|
||||
image_prompt='' # Empty to measure template overhead
|
||||
)
|
||||
@@ -425,7 +449,7 @@ def process_image_generation_queue(self, image_ids: list, account_id: int = None
|
||||
image_prompt = image_prompt[:max_image_prompt_length - 3] + "..."
|
||||
|
||||
formatted_prompt = image_prompt_template.format(
|
||||
image_type=image_type,
|
||||
image_type=style_description, # Use full style description instead of raw value
|
||||
post_title=post_title,
|
||||
image_prompt=image_prompt
|
||||
)
|
||||
@@ -510,6 +534,21 @@ def process_image_generation_queue(self, image_ids: list, account_id: int = None
|
||||
else: # desktop or other (legacy)
|
||||
image_size = in_article_square_size # Default to square
|
||||
|
||||
# For DALL-E, convert image_type to style parameter
|
||||
# image_type is from user settings (e.g., 'vivid', 'natural', 'realistic')
|
||||
# DALL-E accepts 'vivid' or 'natural' - map accordingly
|
||||
dalle_style = None
|
||||
if provider == 'openai':
|
||||
# Map image_type to DALL-E style
|
||||
# 'natural' = more realistic photos (default)
|
||||
# 'vivid' = hyper-real, dramatic images
|
||||
if image_type in ['vivid']:
|
||||
dalle_style = 'vivid'
|
||||
else:
|
||||
# Default to 'natural' for realistic photos
|
||||
dalle_style = 'natural'
|
||||
logger.info(f"[process_image_generation_queue] DALL-E style: {dalle_style} (from image_type: {image_type})")
|
||||
|
||||
result = ai_core.generate_image(
|
||||
prompt=formatted_prompt,
|
||||
provider=provider,
|
||||
@@ -517,7 +556,8 @@ def process_image_generation_queue(self, image_ids: list, account_id: int = None
|
||||
size=image_size,
|
||||
api_key=api_key,
|
||||
negative_prompt=negative_prompt,
|
||||
function_name='generate_images_from_prompts'
|
||||
function_name='generate_images_from_prompts',
|
||||
style=dalle_style
|
||||
)
|
||||
|
||||
# Update progress: Image generation complete (50%)
|
||||
|
||||
Reference in New Issue
Block a user