set-cookie: dmsess=s%3AEsF23MoyDEq7tTWQM8KfA_wjKkSrOFwU.2fjJTfDP%2FT2BeA5DFenwOH4t8XzqZsbSc6M2mZwS%2BWg;

Domain=.deepmotion.com; Path=/; Expires=Mon, 03 Aug 2020 13:36:26 GMT; HttpOnly

(Note: dmsess is the session cookie. This cookie needs to be sent in all subsequent REST API calls.

Sample Request Header for other API calls:

cookie:dmsess=s%3AEsF23MoyDEq7tTWQM8KfA_wjKkSrOFwU.2fjJTfDP%2FT2BeA5DFenwOH4t8XzqZsbSc6M2mZwS%2BWg)

{"credits":<value>,"subscription":{"name":<value>,"credits":<value>,"featureLimits":{"maxVariantsGeneration":<value>},"currentPeriod":{"start":<value>,"end":<value>}}}

{

“params": [<params>, ...]

}.

<processor> specifies which processor to use to process the job, must be one of the following:

<params> specifies additional parameters that will be passed to the specified processor.

For the **text2motion **processor, here are the parameters for a regular job.

"params":

[

“prompt=<value>”,

"model=<value>”,

"requestedAnimationDuration=<value>”,

“dis=<value>”,

“footLockingMode=<value>”,

“poseFilteringStrength=<value>”,

“rootAtOrigin=<value>”

]

And here are the additional parameters for an inpainting job

"params":

[

“t2m_rid=<value>”,

“variant_id=<value>”,

“inPaintingRequest={ “prompt” :<value>, “intervals” : [ { “start” :<value>, “end”:<value> } ] }”

]

prompt

A detailed text prompt to generate motion with. For inpainting jobs, the prompt parameter should be included inside the inPaintingRequest parameter, instead of as a standalone parameter in a regular job.

model

The 3d model used for showcasing the generated motion/animation

dis (optional)

This parameter influences motion generation to improve it in some cases like inter body parts penetration etc, but may cause side effects in animation quality sometimes. By default simulation is turned on. Use dis=s to turn off the simulation.

footLockingMode (optional)

• This parameter value can be one of the below:
  • auto : default mode, automatic switching between locking and gliding modes of the foot, recommended for general cases
  • always : forced foot locking all the time. only used when Auto mode can not remove all the foot gliding unsired
  • never : forced to disable foot locking and character grounding. used when the motion is completely in the air or in the water and therefore neither foot locking nor character grounding is needed.
  • grounding : forced disabling foot locking, however character is still grounded. Only used when Auto mode prevents the desired foot gliding (i.e. during shuffling dances) in the motion or locks the foot for too long on the ground during fast and short foot/ground contacts (i.e. during sprints or jumps.)

poseFilteringStrength (optional)

• Applies an advanced AI filter that helps remove jitter and produce smoother animations though may result in lower animation accuracy for certain frames or sequences
• Default value is 0.0 and range is 0.0 - 1.0

rootAtOrigin (optional)

• Place a root joint at the origin of the output character. This is helpful in some cases, for example, for UE4 retargeting.
• Default value is 0 and value can be either 0 or 1

requestedAnimationDuration (optional)

• Float, request animation generation duration in seconds

t2m_rid

• It is a previous text2motion job request id from which to generate inpainting jobs

variant_id

• It is a specific variant animation id from the above previously generated text2motion job

inPaintingRequest

• A json string, containing inpainting prompt and intervals in frame numbers.

For the **render **processor, here are the parameters.

Only two required parameters are t2m_rid which is a previous text2motion job request id & variant_id which is a specific variant animation from the above previously generated text2motion job. Other optional parameters are below:

*To replace the default background with a solid color (for green screening etc.)

bgColor=0,177,64,0 (RGBA color code in the range of 0-255 for each channel, please note, the last channel (alpha) value is not in effect )

*To set a studio like 3d background with a solid color tint

backdrop=studio

**bgColor=0,177,64,0 **

Also, List of supported 2D backdrops:

• Checker_Cloudscape
• Golden_Age_Glitz
• Inferno_Stage
• Jungle_Grove
• Mystic_Brick
• Mystic_Concrete
• Rainbow_Spotlight
• Red_Carpet
• Retro_Revue
• Sakura_Dreamscape
• Spotlit_Stage
• Timbered_Retreat
• Urban_Rooftop

*To enable character shadow

**shadow=1 ** (not applicable for 2D backdrop)

*camMode

values are below. Default is 0

0 (Cinematic) The character is kept in the center of the frame

1 (Fixed) Camera will stay fixed relative to the background

2 (Face) Camera keeps the torso and face in the center of frame

*camHorizontalAngle

Camera horizontal angle in degrees, where zero means forward camera

• Default value is 0 and range is -90 - +90

Response

{

"rid": <request id>

}

GET {host}/job/v1/status/rid1,rid2,..,rid

Use comma (‘,’) to separate multiple request ids if retrieving status for more than 1 request.

{

"count": <number of records in status array>,

"status": [

<status>,

… …

]

}

Each element in status array is a JSON object:

{

"rid": <request id>,

"status": <status name>,

"details": <status details, see below>,

“positionInQueue”: <position in the queue for only PROGRESS status>

}

<status name> is one of the following case sensitive values:

<status details> for PROGRESS:

{

“step”: <current step>,

“total”: <expected total number of steps>

}

<status details> for SUCCESS:

{

“In”: <original video file>,

“out”: <processed video file>

}

<status details> for RETRY and FAILURE include last error message. Currently the format is:

{

“exc_message”: <exception message, if any>,

“exc_type”: <exception type, if any>

}

But please note the format may change if we decide to mask error information (or pass more information) to client applications.

GET {host}/job/v1//download/rid1,rid2,...,rid?variant_id=2,3,...,n

Use comma (‘,’) to separate request ids if retrieving download URLs for multiple processing requests.

A query parameter called variant_id to be able to download specific variant animation related resources. Use comma (‘,’) to separate variant_id values which correspond to individual rid of multiple job requests

{

“count”: <number of records in links array>,

“links”: [

<link>,

… ...

]

}

Each element in links array is a JSON object:

{

“rid”: <request id>,

“parameters”: <input params of the job>,

“renderJobList” (available only for text2motion jobs):<value>

“variantDownloadStatus”: <boolean flag, false when variant id is wrong or user doesn’t has sufficient balance to download this variant data>

“urls”: [

{

“name”: <name of the downloadable item>

“files”: <links of the files by extension> [

{ <file type>: <URL to download the corresponding file>},

{<file type>: <URL to download the corresponding file>}

]

}

]

}

Please note that if the specified request has not finished yet or has failed, the response will not include any download urls, and the link object will look like:

{

“rid”: “1234567890”

}

Note: failed jobs and old jobs may be removed by system after a predefined retention period

GET {host}/job/v1/list/status1,...,status/processor

Client can specify one or multiple status value(s) along with processor id to retrieve only request ids with the same status value(s). For example, GET /list/PROGRESS/text2motion will only return list of text2motion requests that are still being processed

{

"count": <number of records in the rids array>,

"list": [

{

"rid": <job request id>,

"processingInfo":<detail timing of the processing>,

"processor":<processor id>,

"parameters":<job input parameters>,

“variants” (available only for text2motion jobs): <{“variant_id”:{“download”:<download flag>}}>

“chargedAmount”: <credits used for this job>

"status": <status of the job>,

"ctime": <creation time,milliseconds since epoch>,

"mtime": <last modification time, milliseconds since epoch>

},

... ...

]

}

{

“result”: true

}

<name>: base name of the files (without extension) (optional)

<modelExt>: file extension of the model file. Example: fbx (optional)

<thumbExt>: file extension of the thumb file. Example: jpg (optional)

<resumable>: 0 or 1(default) returns resumable or regular signed url (optional)

{

“modelUrl”: signed url

“thumbUrl”: signed url

}

After retrieving the urls, actual model & thumbnail upload are required to that storage urls. If ’resumable’ option is set in the request, we need one POST and one subsequent PUT request for each signed url, otherwise a single PUT request will do the job per url.

POST request to url:

<x-goog-resumable>: start (set in the request header)

<location>: resumable url (set in the response header by server)

PUT request to resumable url location/url:

attach raw bytes of the model or thumbnail file in the request body.

<modelUrl>: model url returned from API 1 (optional if <modelId> is provided)

<modelName>: model name (optional)

<thumbUrl>: thumbnail url returned from API 1 (optional)

<modelId>: model id to update existing model info (name or thumb) (optional if <modelUrl> is provided)

<createThumb>: 0 (default) or 1, indicate if the thumbnail of the model needs to be generated (optional)

{

“modelId”: <Unique model id that can be passed to text2motion process API>,

“faceDataType”:<if the model supports facial rig>,

“handDataType”:<if the model supports hand/finger rig>

}

<modelId>: existing model id (optional)

<searchToken>: for example search by model name (optional)

<stockModel>: = When this parameter is supplied, all stock models (including deepmotion & roblox) will return in api response along with the account's custom models. Beside that, each model details now include a platform field which can be one of the below values : custom, deepmotion, roblox . (optional)

[

{

“Id: Unique model id that can be passed to video process API

“name”: name of the model

“thumb”: url of the thumbnail if exist

“glb”: url of glb format of the character

“rigId”: rigTemplate id with which this model is associated with

“ctime”: creation timestamp

“mtime”: modification timestamp

“platform”: platform of the model

}

]

{

“count”: number of models that have been deleted. Currently only one model can be deleted at a time.

}