github/AutoGPT
1
1
Fork 0
mirror of https://github.com/Significant-Gravitas/AutoGPT.git synced 2026-01-09 23:28:07 -05:00
Code Issues Packages Projects Releases Wiki Activity

feat(blocks): Add more Revid.ai media generation blocks (#9931)

Browse Source 
<html><head></head><body><h3>Why these changes are needed 🧐</h3>
<p>Revid.ai offers several specialised, undocumented rendering flows
beyond the basic “text-to-video” endpoint our platform already
supported.
to:</p>
<ol>
<li>
<p><strong>Generate ads</strong> from copy plus product images
(30-second vertical spots).</p>
</li>
<li>
<p><strong>Turn a single creative prompt</strong> into a fully
AI-generated video (no multi-line script).</p>
</li>
<li>
<p><strong>Transform a screenshot into a narrated, avatar-driven
clip</strong>, ideal for product-led demos.</p>
</li>
</ol>
<p>Without first-class blocks for these flows, users were forced to drop
to raw HTTP nodes, losing schema validation, test mocks and credential
management.</p>
<h3>Changes 🏗️</h3>

Added new category to ``BlockCategory`` in ``block.py`` for ``MARKETING
= "Block that helps with marketing"``

Area | Change | Notes
-- | -- | --
ai_shortform_video_block.py | Refactored out a shared _RevidMixin
(webhook + polling helpers). | Keeps DRY across new blocks.
  | Added AudioTrack.DONT_STOP_ME_ABSTRACT_FUTURE_BASS and Voice.EVA
enum members. | Required by Revid sample payloads.
  | AIAdMakerVideoCreatorBlock | Implements ai-ad-generator flow;
supports optional input_media_urls, target_duration,
use_only_provided_media.
  | AIPromptToVideoCreatorBlock | Implements prompt-to-video flow with
prompt_target_duration.
  | AIScreenshotToVideoAdBlock | Implements screenshot-to-video-ad flow
(avatar narration, BG removal).
  | Added full pydantic schemas, test stubs & mock hooks for each new
block. | Ensures unit tests pass and blocks appear in UI.


<p>No existing functionality was removed; current <code
inline="">AIShortformVideoCreatorBlock</code> is untouched apart from
enum imports.</p></body></html>

### Checklist 📋

#### For code changes:
- [x] I have clearly listed my changes in the PR description
- [x] I have made a test plan
- [x] I have tested my changes according to the test plan:
  <!-- Put your test plan here: -->
- [x] use the ``AI ShortForm Video Creator`` block to generate a video
and it will work
- [x] same with `` ai ad maker video creator`` block test it and it
should work
  - [x] and test ``ai screenshot to video ad`` block it should work

---------

Co-authored-by: Bently <Github@bentlybro.com>
This commit is contained in:
Toran Bruce Richards
2025-06-27 07:39:24 +01:00
committed by GitHub
parent 7f7c387156
commit 77e99e9739
2 changed files with 415 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

457
autogpt_platform/backend/backend/blocks/ai_shortform_video_block.py
View File

@@ -53,6 +53,7 @@ class AudioTrack(str, Enum):
REFRESHER = ("Refresher",)
TOURIST = ("Tourist",)
TWIN_TYCHES = ("Twin Tyches",)
DONT_STOP_ME_ABSTRACT_FUTURE_BASS = ("Dont Stop Me Abstract Future Bass",)
@property
def audio_url(self):
@@ -78,6 +79,7 @@ class AudioTrack(str, Enum):
AudioTrack.REFRESHER: "https://cdn.tfrv.xyz/audio/refresher.mp3",
AudioTrack.TOURIST: "https://cdn.tfrv.xyz/audio/tourist.mp3",
AudioTrack.TWIN_TYCHES: "https://cdn.tfrv.xyz/audio/twin-tynches.mp3",
AudioTrack.DONT_STOP_ME_ABSTRACT_FUTURE_BASS: "https://cdn.revid.ai/audio/_dont-stop-me-abstract-future-bass.mp3",
}
return audio_urls[self]
@@ -105,6 +107,7 @@ class GenerationPreset(str, Enum):
MOVIE = ("Movie",)
STYLIZED_ILLUSTRATION = ("Stylized Illustration",)
MANGA = ("Manga",)
DEFAULT = ("DEFAULT",)
class Voice(str, Enum):
@@ -114,6 +117,7 @@ class Voice(str, Enum):
JESSICA = "Jessica"
CHARLOTTE = "Charlotte"
CALLUM = "Callum"
EVA = "Eva"
@property
def voice_id(self):
@@ -124,6 +128,7 @@ class Voice(str, Enum):
Voice.JESSICA: "cgSgspJ2msm6clMCkdW9",
Voice.CHARLOTTE: "XB0fDUnXU5powFXDhCwa",
Voice.CALLUM: "N2lVS1w4EtoT3dr4eOWO",
Voice.EVA: "FGY2WhTYpPnrIDTdsKH5",
}
return voice_id_map[self]
@@ -141,6 +146,8 @@ logger = logging.getLogger(__name__)
class AIShortformVideoCreatorBlock(Block):
"""Creates a short‑form text‑to‑video clip using stock or AI imagery."""
class Input(BlockSchema):
credentials: CredentialsMetaInput[
Literal[ProviderName.REVID], Literal["api_key"]
@@ -184,40 +191,8 @@ class AIShortformVideoCreatorBlock(Block):
video_url: str = SchemaField(description="The URL of the created video")
error: str = SchemaField(description="Error message if the request failed")
def __init__(self):
super().__init__(
id="361697fb-0c4f-4feb-aed3-8320c88c771b",
description="Creates a shortform video using revid.ai",
categories={BlockCategory.SOCIAL, BlockCategory.AI},
input_schema=AIShortformVideoCreatorBlock.Input,
output_schema=AIShortformVideoCreatorBlock.Output,
test_input={
"credentials": TEST_CREDENTIALS_INPUT,
"script": "[close-up of a cat] Meow!",
"ratio": "9 / 16",
"resolution": "720p",
"frame_rate": 60,
"generation_preset": GenerationPreset.LEONARDO,
"background_music": AudioTrack.HIGHWAY_NOCTURNE,
"voice": Voice.LILY,
"video_style": VisualMediaType.STOCK_VIDEOS,
},
test_output=(
"video_url",
"https://example.com/video.mp4",
),
test_mock={
"create_webhook": lambda: (
"test_uuid",
"https://webhook.site/test_uuid",
),
"create_video": lambda api_key, payload: {"pid": "test_pid"},
"wait_for_video": lambda api_key, pid, webhook_token, max_wait_time=1000: "https://example.com/video.mp4",
},
test_credentials=TEST_CREDENTIALS,
)
async def create_webhook(self):
async def create_webhook(self) -> tuple[str, str]:
"""Create a new webhook URL for receiving notifications."""
url = "https://webhook.site/token"
headers = {"Accept": "application/json", "Content-Type": "application/json"}
response = await Requests().post(url, headers=headers)
@@ -225,6 +200,7 @@ class AIShortformVideoCreatorBlock(Block):
return webhook_data["uuid"], f"https://webhook.site/{webhook_data['uuid']}"
async def create_video(self, api_key: SecretStr, payload: dict) -> dict:
"""Create a video using the Revid API."""
url = "https://www.revid.ai/api/public/v2/render"
headers = {"key": api_key.get_secret_value()}
response = await Requests().post(url, json=payload, headers=headers)
@@ -234,6 +210,7 @@ class AIShortformVideoCreatorBlock(Block):
return response.json()
async def check_video_status(self, api_key: SecretStr, pid: str) -> dict:
"""Check the status of a video creation job."""
url = f"https://www.revid.ai/api/public/v2/status?pid={pid}"
headers = {"key": api_key.get_secret_value()}
response = await Requests().get(url, headers=headers)
@@ -243,9 +220,9 @@ class AIShortformVideoCreatorBlock(Block):
self,
api_key: SecretStr,
pid: str,
webhook_token: str,
max_wait_time: int = 1000,
) -> str:
"""Wait for video creation to complete and return the video URL."""
start_time = time.time()
while time.time() - start_time < max_wait_time:
status = await self.check_video_status(api_key, pid)
@@ -266,6 +243,40 @@ class AIShortformVideoCreatorBlock(Block):
logger.error("Video creation timed out")
raise TimeoutError("Video creation timed out")
def __init__(self):
super().__init__(
id="361697fb-0c4f-4feb-aed3-8320c88c771b",
description="Creates a shortform video using revid.ai",
categories={BlockCategory.SOCIAL, BlockCategory.AI},
input_schema=AIShortformVideoCreatorBlock.Input,
output_schema=AIShortformVideoCreatorBlock.Output,
test_input={
"credentials": TEST_CREDENTIALS_INPUT,
"script": "[close-up of a cat] Meow!",
"ratio": "9 / 16",
"resolution": "720p",
"frame_rate": 60,
"generation_preset": GenerationPreset.LEONARDO,
"background_music": AudioTrack.HIGHWAY_NOCTURNE,
"voice": Voice.LILY,
"video_style": VisualMediaType.STOCK_VIDEOS,
},
test_output=("video_url", "https://example.com/video.mp4"),
test_mock={
"create_webhook": lambda *args, **kwargs: (
"test_uuid",
"https://webhook.site/test_uuid",
),
"create_video": lambda *args, **kwargs: {"pid": "test_pid"},
"check_video_status": lambda *args, **kwargs: {
"status": "ready",
"videoUrl": "https://example.com/video.mp4",
},
"wait_for_video": lambda *args, **kwargs: "https://example.com/video.mp4",
},
test_credentials=TEST_CREDENTIALS,
)
async def run(
self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs
) -> BlockOutput:
@@ -273,20 +284,18 @@ class AIShortformVideoCreatorBlock(Block):
webhook_token, webhook_url = await self.create_webhook()
logger.debug(f"Webhook URL: {webhook_url}")
audio_url = input_data.background_music.audio_url
payload = {
"frameRate": input_data.frame_rate,
"resolution": input_data.resolution,
"frameDurationMultiplier": 18,
"webhook": webhook_url,
"webhook": None,
"creationParams": {
"mediaType": input_data.video_style,
"captionPresetName": "Wrap 1",
"selectedVoice": input_data.voice.voice_id,
"hasEnhancedGeneration": True,
"generationPreset": input_data.generation_preset.name,
"selectedAudio": input_data.background_music,
"selectedAudio": input_data.background_music.value,
"origin": "/create",
"inputText": input_data.script,
"flowType": "text-to-video",
@@ -302,7 +311,7 @@ class AIShortformVideoCreatorBlock(Block):
"selectedStoryStyle": {"value": "custom", "label": "Custom"},
"hasToGenerateVideos": input_data.video_style
!= VisualMediaType.STOCK_VIDEOS,
"audioUrl": audio_url,
"audioUrl": input_data.background_music.audio_url,
},
}
@@ -319,8 +328,370 @@ class AIShortformVideoCreatorBlock(Block):
logger.debug(
f"Video created with project ID: {pid}. Waiting for completion..."
)
video_url = await self.wait_for_video(
credentials.api_key, pid, webhook_token
)
video_url = await self.wait_for_video(credentials.api_key, pid)
logger.debug(f"Video ready: {video_url}")
yield "video_url", video_url
class AIAdMakerVideoCreatorBlock(Block):
"""Generates a 30‑second vertical AI advert using optional user‑supplied imagery."""
class Input(BlockSchema):
credentials: CredentialsMetaInput[
Literal[ProviderName.REVID], Literal["api_key"]
] = CredentialsField(
description="Credentials for Revid.ai API access.",
)
script: str = SchemaField(
description="Short advertising copy. Line breaks create new scenes.",
placeholder="Introducing Foobar – [show product photo] the gadget that does it all.",
)
ratio: str = SchemaField(description="Aspect ratio", default="9 / 16")
target_duration: int = SchemaField(
description="Desired length of the ad in seconds.", default=30
)
voice: Voice = SchemaField(
description="Narration voice", default=Voice.EVA, placeholder=Voice.EVA
)
background_music: AudioTrack = SchemaField(
description="Background track",
default=AudioTrack.DONT_STOP_ME_ABSTRACT_FUTURE_BASS,
)
input_media_urls: list[str] = SchemaField(
description="List of image URLs to feature in the advert.", default=[]
)
use_only_provided_media: bool = SchemaField(
description="Restrict visuals to supplied images only.", default=True
)
class Output(BlockSchema):
video_url: str = SchemaField(description="URL of the finished advert")
error: str = SchemaField(description="Error message on failure")
async def create_webhook(self) -> tuple[str, str]:
"""Create a new webhook URL for receiving notifications."""
url = "https://webhook.site/token"
headers = {"Accept": "application/json", "Content-Type": "application/json"}
response = await Requests().post(url, headers=headers)
webhook_data = response.json()
return webhook_data["uuid"], f"https://webhook.site/{webhook_data['uuid']}"
async def create_video(self, api_key: SecretStr, payload: dict) -> dict:
"""Create a video using the Revid API."""
url = "https://www.revid.ai/api/public/v2/render"
headers = {"key": api_key.get_secret_value()}
response = await Requests().post(url, json=payload, headers=headers)
logger.debug(
f"API Response Status Code: {response.status}, Content: {response.text}"
)
return response.json()
async def check_video_status(self, api_key: SecretStr, pid: str) -> dict:
"""Check the status of a video creation job."""
url = f"https://www.revid.ai/api/public/v2/status?pid={pid}"
headers = {"key": api_key.get_secret_value()}
response = await Requests().get(url, headers=headers)
return response.json()
async def wait_for_video(
self,
api_key: SecretStr,
pid: str,
max_wait_time: int = 1000,
) -> str:
"""Wait for video creation to complete and return the video URL."""
start_time = time.time()
while time.time() - start_time < max_wait_time:
status = await self.check_video_status(api_key, pid)
logger.debug(f"Video status: {status}")
if status.get("status") == "ready" and "videoUrl" in status:
return status["videoUrl"]
elif status.get("status") == "error":
error_message = status.get("error", "Unknown error occurred")
logger.error(f"Video creation failed: {error_message}")
raise ValueError(f"Video creation failed: {error_message}")
elif status.get("status") in ["FAILED", "CANCELED"]:
logger.error(f"Video creation failed: {status.get('message')}")
raise ValueError(f"Video creation failed: {status.get('message')}")
await asyncio.sleep(10)
logger.error("Video creation timed out")
raise TimeoutError("Video creation timed out")
def __init__(self):
super().__init__(
id="58bd2a19-115d-4fd1-8ca4-13b9e37fa6a0",
description="Creates an AI‑generated 30‑second advert (text + images)",
categories={BlockCategory.MARKETING, BlockCategory.AI},
input_schema=AIAdMakerVideoCreatorBlock.Input,
output_schema=AIAdMakerVideoCreatorBlock.Output,
test_input={
"credentials": TEST_CREDENTIALS_INPUT,
"script": "Test product launch!",
"input_media_urls": [
"https://cdn.revid.ai/uploads/1747076315114-image.png",
],
},
test_output=("video_url", "https://example.com/ad.mp4"),
test_mock={
"create_webhook": lambda *args, **kwargs: (
"test_uuid",
"https://webhook.site/test_uuid",
),
"create_video": lambda *args, **kwargs: {"pid": "test_pid"},
"check_video_status": lambda *args, **kwargs: {
"status": "ready",
"videoUrl": "https://example.com/ad.mp4",
},
"wait_for_video": lambda *args, **kwargs: "https://example.com/ad.mp4",
},
test_credentials=TEST_CREDENTIALS,
)
async def run(self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs):
webhook_token, webhook_url = await self.create_webhook()
payload = {
"webhook": webhook_url,
"creationParams": {
"targetDuration": input_data.target_duration,
"ratio": input_data.ratio,
"mediaType": "aiVideo",
"inputText": input_data.script,
"flowType": "text-to-video",
"slug": "ai-ad-generator",
"slugNew": "",
"isCopiedFrom": False,
"hasToGenerateVoice": True,
"hasToTranscript": False,
"hasToSearchMedia": True,
"hasAvatar": False,
"hasWebsiteRecorder": False,
"hasTextSmallAtBottom": False,
"selectedAudio": input_data.background_music.value,
"selectedVoice": input_data.voice.voice_id,
"selectedAvatar": "https://cdn.revid.ai/avatars/young-woman.mp4",
"selectedAvatarType": "video/mp4",
"websiteToRecord": "",
"hasToGenerateCover": True,
"nbGenerations": 1,
"disableCaptions": False,
"mediaMultiplier": "medium",
"characters": [],
"captionPresetName": "Revid",
"sourceType": "contentScraping",
"selectedStoryStyle": {"value": "custom", "label": "General"},
"generationPreset": "DEFAULT",
"hasToGenerateMusic": False,
"isOptimizedForChinese": False,
"generationUserPrompt": "",
"enableNsfwFilter": False,
"addStickers": False,
"typeMovingImageAnim": "dynamic",
"hasToGenerateSoundEffects": False,
"forceModelType": "gpt-image-1",
"selectedCharacters": [],
"lang": "",
"voiceSpeed": 1,
"disableAudio": False,
"disableVoice": False,
"useOnlyProvidedMedia": input_data.use_only_provided_media,
"imageGenerationModel": "ultra",
"videoGenerationModel": "pro",
"hasEnhancedGeneration": True,
"hasEnhancedGenerationPro": True,
"inputMedias": [
{"url": url, "title": "", "type": "image"}
for url in input_data.input_media_urls
],
"hasToGenerateVideos": True,
"audioUrl": input_data.background_music.audio_url,
"watermark": None,
},
}
response = await self.create_video(credentials.api_key, payload)
pid = response.get("pid")
if not pid:
raise RuntimeError("Failed to create video: No project ID returned")
video_url = await self.wait_for_video(credentials.api_key, pid)
yield "video_url", video_url
class AIScreenshotToVideoAdBlock(Block):
"""Creates an advert where the supplied screenshot is narrated by an AI avatar."""
class Input(BlockSchema):
credentials: CredentialsMetaInput[
Literal[ProviderName.REVID], Literal["api_key"]
] = CredentialsField(description="Revid.ai API key")
script: str = SchemaField(
description="Narration that will accompany the screenshot.",
placeholder="Check out these amazing stats!",
)
screenshot_url: str = SchemaField(
description="Screenshot or image URL to showcase."
)
ratio: str = SchemaField(default="9 / 16")
target_duration: int = SchemaField(default=30)
voice: Voice = SchemaField(default=Voice.EVA)
background_music: AudioTrack = SchemaField(
default=AudioTrack.DONT_STOP_ME_ABSTRACT_FUTURE_BASS
)
class Output(BlockSchema):
video_url: str = SchemaField(description="Rendered video URL")
error: str = SchemaField(description="Error, if encountered")
async def create_webhook(self) -> tuple[str, str]:
"""Create a new webhook URL for receiving notifications."""
url = "https://webhook.site/token"
headers = {"Accept": "application/json", "Content-Type": "application/json"}
response = await Requests().post(url, headers=headers)
webhook_data = response.json()
return webhook_data["uuid"], f"https://webhook.site/{webhook_data['uuid']}"
async def create_video(self, api_key: SecretStr, payload: dict) -> dict:
"""Create a video using the Revid API."""
url = "https://www.revid.ai/api/public/v2/render"
headers = {"key": api_key.get_secret_value()}
response = await Requests().post(url, json=payload, headers=headers)
logger.debug(
f"API Response Status Code: {response.status}, Content: {response.text}"
)
return response.json()
async def check_video_status(self, api_key: SecretStr, pid: str) -> dict:
"""Check the status of a video creation job."""
url = f"https://www.revid.ai/api/public/v2/status?pid={pid}"
headers = {"key": api_key.get_secret_value()}
response = await Requests().get(url, headers=headers)
return response.json()
async def wait_for_video(
self,
api_key: SecretStr,
pid: str,
max_wait_time: int = 1000,
) -> str:
"""Wait for video creation to complete and return the video URL."""
start_time = time.time()
while time.time() - start_time < max_wait_time:
status = await self.check_video_status(api_key, pid)
logger.debug(f"Video status: {status}")
if status.get("status") == "ready" and "videoUrl" in status:
return status["videoUrl"]
elif status.get("status") == "error":
error_message = status.get("error", "Unknown error occurred")
logger.error(f"Video creation failed: {error_message}")
raise ValueError(f"Video creation failed: {error_message}")
elif status.get("status") in ["FAILED", "CANCELED"]:
logger.error(f"Video creation failed: {status.get('message')}")
raise ValueError(f"Video creation failed: {status.get('message')}")
await asyncio.sleep(10)
logger.error("Video creation timed out")
raise TimeoutError("Video creation timed out")
def __init__(self):
super().__init__(
id="0f3e4635-e810-43d9-9e81-49e6f4e83b7c",
description="Turns a screenshot into an engaging, avatar‑narrated video advert.",
categories={BlockCategory.AI, BlockCategory.MARKETING},
input_schema=AIScreenshotToVideoAdBlock.Input,
output_schema=AIScreenshotToVideoAdBlock.Output,
test_input={
"credentials": TEST_CREDENTIALS_INPUT,
"script": "Amazing numbers!",
"screenshot_url": "https://cdn.revid.ai/uploads/1747080376028-image.png",
},
test_output=("video_url", "https://example.com/screenshot.mp4"),
test_mock={
"create_webhook": lambda *args, **kwargs: (
"test_uuid",
"https://webhook.site/test_uuid",
),
"create_video": lambda *args, **kwargs: {"pid": "test_pid"},
"check_video_status": lambda *args, **kwargs: {
"status": "ready",
"videoUrl": "https://example.com/screenshot.mp4",
},
"wait_for_video": lambda *args, **kwargs: "https://example.com/screenshot.mp4",
},
test_credentials=TEST_CREDENTIALS,
)
async def run(self, input_data: Input, *, credentials: APIKeyCredentials, **kwargs):
webhook_token, webhook_url = await self.create_webhook()
payload = {
"webhook": webhook_url,
"creationParams": {
"targetDuration": input_data.target_duration,
"ratio": input_data.ratio,
"mediaType": "aiVideo",
"hasAvatar": True,
"removeAvatarBackground": True,
"inputText": input_data.script,
"flowType": "text-to-video",
"slug": "ai-ad-generator",
"slugNew": "screenshot-to-video-ad",
"isCopiedFrom": "ai-ad-generator",
"hasToGenerateVoice": True,
"hasToTranscript": False,
"hasToSearchMedia": True,
"hasWebsiteRecorder": False,
"hasTextSmallAtBottom": False,
"selectedAudio": input_data.background_music.value,
"selectedVoice": input_data.voice.voice_id,
"selectedAvatar": "https://cdn.revid.ai/avatars/young-woman.mp4",
"selectedAvatarType": "video/mp4",
"websiteToRecord": "",
"hasToGenerateCover": True,
"nbGenerations": 1,
"disableCaptions": False,
"mediaMultiplier": "medium",
"characters": [],
"captionPresetName": "Revid",
"sourceType": "contentScraping",
"selectedStoryStyle": {"value": "custom", "label": "General"},
"generationPreset": "DEFAULT",
"hasToGenerateMusic": False,
"isOptimizedForChinese": False,
"generationUserPrompt": "",
"enableNsfwFilter": False,
"addStickers": False,
"typeMovingImageAnim": "dynamic",
"hasToGenerateSoundEffects": False,
"forceModelType": "gpt-image-1",
"selectedCharacters": [],
"lang": "",
"voiceSpeed": 1,
"disableAudio": False,
"disableVoice": False,
"useOnlyProvidedMedia": True,
"imageGenerationModel": "ultra",
"videoGenerationModel": "ultra",
"hasEnhancedGeneration": True,
"hasEnhancedGenerationPro": True,
"inputMedias": [
{"url": input_data.screenshot_url, "title": "", "type": "image"}
],
"hasToGenerateVideos": True,
"audioUrl": input_data.background_music.audio_url,
"watermark": None,
},
}
response = await self.create_video(credentials.api_key, payload)
pid = response.get("pid")
if not pid:
raise RuntimeError("Failed to create video: No project ID returned")
video_url = await self.wait_for_video(credentials.api_key, pid)
yield "video_url", video_url

1
autogpt_platform/backend/backend/data/block.py
View File

@@ -78,6 +78,7 @@ class BlockCategory(Enum):
PRODUCTIVITY = "Block that helps with productivity"
ISSUE_TRACKING = "Block that helps with issue tracking"
MULTIMEDIA = "Block that interacts with multimedia content"
MARKETING = "Block that helps with marketing"
def dict(self) -> dict[str, str]:
return {"category": self.name, "description": self.value}