a bunch of stuff

app/animations/animation.py

@@ -2,20 +2,53 @@
 from PIL import Image, ImageDraw
 
 from ..plugins import replicate, elevenlabs, s3
-from ..character import EdenCharacter
-from ..utils import combine_speech_video, wrap_text, get_font
+from ..character import Character
+from ..utils import combine_speech_video, wrap_text, get_font, create_dynamic_model
+from ..scenarios.tasks import general_assistant
+from ..models.tasks import SimpleAssistantRequest
+from ..plugins import elevenlabs
+
+
+def select_random_voice(character: Character = None):
+    if not character:
+        return elevenlabs.get_random_voice()
+
+    gender_schema = create_dynamic_model("gender", ["male", "female"])
+
+    prompt = f"What is the most likely gender of the following character, male or female?\n\nName: {character.name}\n\nDescription: {character.identity}"
+
+    request = SimpleAssistantRequest(
+        prompt=prompt,
+        model="gpt-3.5-turbo",
+        params={"temperature": 0.0, "max_tokens": 10},
+        output_schema=gender_schema
+    )
+
+    try:
+        gender = general_assistant(request)
+        voice_id = elevenlabs.get_random_voice(gender=gender)
+    except Exception as e:
+        voice_id = elevenlabs.get_random_voice()
+    finally:
+        return voice_id
 
 
 def talking_head(
-    character: EdenCharacter,
+    character: Character,
     text: str, 
     width: Optional[int] = None,
     height: Optional[int] = None
 ) -> str:
+    if character.voice:
+        voice_id = character.voice
+    else:
+        voice_id = select_random_voice(character)
+
     audio_bytes = elevenlabs.tts(
         text, 
-        voice=character.voice
+        voice=voice_id
     )
+
     audio_url = s3.upload(audio_bytes, "mp3")
     output_url, thumbnail_url = replicate.wav2lip(
         face_url=character.image,
@@ -29,15 +62,22 @@ def talking_head(
 
 
 def screenplay_clip(
-    character: EdenCharacter,
+    character: Optional[Character],
     speech: str,
     image_text: str,
     width: Optional[int] = None,
     height: Optional[int] = None
 ) -> str:
+    if not character:
+        voice_id = select_random_voice()
+    else:
+        if character.voice:
+            voice_id = character.voice
+        else:
+            voice_id = select_random_voice(character)
     audio_bytes = elevenlabs.tts(
         speech, 
-        voice=character.voice
+        voice=voice_id
     )
     audio_url = s3.upload(audio_bytes, "mp3")
     video_url, thumbnail_url = replicate.txt2vid(
@@ -57,7 +97,7 @@ def comic_strip(
     caption_padding_top: int = 10,
     line_spacing: int = 1.3,
     font_size: int = 48,
-    font_ttf: str = 'Roboto-Regular.ttf'
+    font_ttf: str = 'Raleway-Light.ttf'
 ):
     font = get_font(font_ttf, font_size)
     num_panels = len(images)
@@ -119,19 +159,19 @@ def comic_strip(
 def poster(
     image: Image.Image,
     caption: str,
-    margin: int = 10,
+    margin: int = 32,
     caption_padding_top: int = 10,
     line_spacing: int = 1.3,
-    font_size: int = 48,
-    font_ttf: str = 'Roboto-Regular.ttf',
+    font_size: int = 36,
+    font_ttf: str = 'Raleway-Light.ttf',
     shadow_offset: tuple = (1, 1.4),
     font_color: str = '#e7e7e7',
     shadow_color: str = '#d3d3d3'
 ):
     font = get_font(font_ttf, font_size)
     width, height = image.size
 
-    draw = ImageDraw.Draw(Image.new('RGB', (width, height)))  # Temporary image for wrapping text
+    draw = ImageDraw.Draw(Image.new('RGB', (width, height)))
     wrapped_caption = wrap_text(draw, caption, font, width - 2 * margin)
     num_lines = len(wrapped_caption)
 
@@ -152,7 +192,7 @@ def poster(
     caption_box = Image.new('RGB', (width, caption_box_height), color='black')
     draw = ImageDraw.Draw(caption_box)
 
-    caption_y = caption_padding_top + int(margin/2)
+    caption_y = caption_padding_top + 0*margin/2
     for line in wrapped_caption:
         draw.text((margin + shadow_offset[0], caption_y + shadow_offset[1]), line, fill=shadow_color, font=font)
         draw.text((margin, caption_y), line, fill=font_color, font=font)

app/animations/dialogue.py

@@ -26,6 +26,7 @@ def animated_dialogue(request: DialogueRequest):
         characters[character_id].image 
         for character_id in request.character_ids
     ]
+    images = list(set(images))
     width, height = utils.calculate_target_dimensions(images, MAX_PIXELS)
 
     def run_talking_head_segment(message, idx):
@@ -61,6 +62,8 @@ def run_talking_head_segment(message, idx):
             os.remove(video_file)
 
     # generate thumbnail
+    print("DO IMAGES")
+    print(images)
     thumbnail = utils.create_dialogue_thumbnail(*images, width, height)
     thumbnail_url = s3.upload(thumbnail, "webp")
 

app/animations/little_martians.py

@@ -29,44 +29,30 @@ def little_martian_poster(request: LittleMartianRequest):
     init_image = random.choice(data['init_images'])
     init_image_strength = random_interval(*data['init_image_strength'])
 
-
-    print("go")
-
-
-    print(lora)
-    print(modifier)
-    print(lora_scale)
-
-
     littlemartian_writer = LLM(
         model=request.model,
         system_message=littlemartians_poster_system.template,
         params=params,
     )
-    print("------")
-    print(request.prompt)
-
-
+
     prompt = littlemartians_poster_prompt.substitute(
         martian = request.martian.value,
         setting = request.setting.value,
         genre = request.genre.value,
         premise = request.prompt,
     )
 
-
-    print(prompt)
+    print("==============")
+    print("PROMPT:", prompt)
 
     result = littlemartian_writer(prompt, output_schema=Poster)
 
-    print(result)
     prompt = result['image']
 
-
     text_input = f'{modifier}, {prompt}'
-    # text_input = f'{modifier}'
 
-    print(text_input)
+    print("RESULT:", prompt)
+    print("text_input:", text_input)
 
     # def run_panel(panel, idx):
     #     # pick lora of character
@@ -94,6 +80,8 @@ def little_martian_poster(request: LittleMartianRequest):
         "n_samples": 1,
     }
 
+
+    print("config")
     print(config)
 
 
@@ -102,9 +90,12 @@ def little_martian_poster(request: LittleMartianRequest):
     print(image_url, thumbnail_url)
 
     caption = result['caption']
+
     print(caption)
     image = utils.download_image(image_url)
     composite_image, thumbnail_image = poster(image, caption)
+
+    print("-========-")
     # results = utils.process_in_parallel(
     #     comic_book['panels'], 
     #     run_panel,
@@ -123,4 +114,16 @@ def little_martian_poster(request: LittleMartianRequest):
     output_url = s3.upload(img_bytes, "jpg")
     thumbnail_url = s3.upload(thumbnail_bytes, "webp")
 
+    # download output_url
+    # martian, setting, genre, pr = request.martian.value, request.setting.value, request.genre.value, request.prompt
+
+    # filename = f'{martian}_{setting}_{genre}_{pr}.jpg'
+
+    # with open(filename, "wb") as f:
+    #     f.write(requests.get(output_url).content)
+
+    # # save config as json file
+    # with open(f"{filename}.json", "w") as f:
+    #     f.write(str(config))
+
     return output_url, thumbnail_url

app/animations/story.py

@@ -4,7 +4,7 @@
 
 from .. import utils
 from ..plugins import replicate, elevenlabs, s3
-from ..character import EdenCharacter
+from ..character import Character, EdenCharacter
 from ..scenarios import story
 from ..models import StoryRequest
 from .animation import screenplay_clip
@@ -15,6 +15,7 @@
 
 def animated_story(request: StoryRequest):
     screenplay = story(request)
+    #screenplay = {'clips': [{'voiceover': 'narrator', 'character': None, 'speech': 'In a quaint ski town replete with snowy peaks and the charm of winter, there lies a forgotten legend eager to reclaim his throne.', 'image_description': 'The sun glints off the snow-capped mountains, as a quaint alpine village stirs to life in the valley below.'}, {'voiceover': 'character', 'character': 'Viper', 'speech': "You see this medal? That's the mark of a champion. Sometimes I miss the cheers, the rush of the wind... the taste of victory.", 'image_description': 'A bronze-hued man, Viper, sits in a cozy ski lodge. Various medals and trophies litter the space around him. His eyes gleam with pride, yet carry a tinge of yearning.'}, {'voiceover': 'narrator', 'character': None, 'speech': 'But life, like the serpentine trails he once conquered, took an unexpected turn when a new star shot across his sky.', 'image_description': 'Laughter echoes through the local gym as a towering woman playfully scores basket after basket on an indoor court.'}, {'voiceover': 'character', 'character': 'Linda', 'speech': "I thought I knew what it was to be at the top, but Linda, she... she's a mountain of her own, you know?", 'image_description': "Viper gazes through the gym's glass doors, admiration painted on his features as he watches a nine-foot-tall woman, Linda, effortlessly dominate the basketball court."}, {'voiceover': 'narrator', 'character': None, 'speech': 'From the frozen slopes to the polished hardwood, it was as if fate had crafted a match not even the sports gods could have predicted.', 'image_description': 'A cosmic dance of stars and snowflakes swirls around the figures of Viper and Linda as they first meet, their differences dwarfed by the evident spark between them.'}, {'voiceover': 'character', 'character': 'Viper', 'speech': "Hey Linda, want to trade some stories? I've got a lifetime on the slopes and you've got... well, a view from the top. Literally.", 'image_description': 'Viper approaches Linda with a mischievous smirk, leaning on a basketball. Linda towers above him, her presence as commanding as her height.'}, {'voiceover': 'character', 'character': 'Linda', 'speech': "I am Linda I am so cool", 'image_description': 'Linda plays basketball at night'}]}
     print(screenplay)
 
     characters = {
@@ -27,20 +28,61 @@ def animated_story(request: StoryRequest):
         for character_id, character in characters.items()
     }
 
-    # images = [
-    #     characters[character_id].image 
-    #     for character_id in request.character_ids
-    # ]
+    print("LETS MAKE ALL THE CHARACTERS")
+    print("LETS MAKE ALL THE CHARACTERS 2")
+    thechars = [clip['character'] for clip in screenplay['clips']]
+    print("LETS MAKE ALL THE CHARACTERS 3")
+    print(thechars)
+    print(character_name_lookup)
+    # if any character is new, assign a random voice
+    for clip in screenplay['clips']:
+        print(" --> ", clip['character'], clip['voiceover'])
+        if clip['character']:
+            character_name = clip['character']
+            print("ceck " , character_name)
+            print("ceck2 " , character_name)
+            print("look here", character_name_lookup)
+            print("look there", characters)
+            if character_name not in character_name_lookup:
+                print("MAKE A NEW CHARACTER", character_name)
+                characters[character_name] = Character(name=character_name)
+                character_name_lookup[character_name] = character_name
+            else:
+                print("CHARACTER IS FOUND", character_name)
+            print("check on voice...", character_name)
+            print("about to chekc ovice")
+            print("is it voice? 0")
+            print(characters)
+            character_id = character_name_lookup[character_name]
+            print("got id", character_id)
+            print(characters[character_id])
+            print("ok 3")
+            #print(characters[character_name])
+            print("ok 4")
+            print("is it voice?", characters[character_id].voice)
+            if not characters[character_id].voice:
+                print("NO VOICE, GET A RANDOM VOICE")
+                voice_id = elevenlabs.get_random_voice()
+                characters[character_id].voice = voice_id
+            print("is it voice now?", characters[character_id].voice)
+        print("------")    
+
+    print("THIS IS THE END!")
+
+
+
 
     # width, height = utils.calculate_target_dimensions(images, MAX_PIXELS)
-    width, height = 1024, 1024
+    width, height = 1792, 1024
 
     def run_story_segment(clip, idx):
         if clip['voiceover'] == 'character':
             character_id = character_name_lookup[clip['character']]
-            character = characters[character_id]
+            # character_id = characters[clip['character']]
+            character = characters.get(character_id)
         else:
             character = characters[request.narrator_id]
+        #print("")
         output_filename, thumbnail_url = screenplay_clip(
             character,
             clip['speech'],