Autodesk has unveiled Flame® 2021, introducing new features aimed at innovating and accelerating creative workflows for VFX, color grading, look development, and editorial finishing. Flame 2021 increases workflow flexibility for artists, expands A.I. capabilities with new machine learning-powered human face segmentation, and simplifies finishing for streaming services with new functionality for Dolby Vision HDR authoring and display. By popular demand, the release also adds a new GPU-accelerated Physical Defocus effect as well as finishing enhancements to make it easier to adjust looks across many shots, share updates with clients, and create beautiful work, faster.“Flame allows artists to create at the speed of thought – whether iterating quickly for client reviews, adding finishing touches to shots, or creating beautiful content that meets modern streaming standards,” said Will Harris, Flame Family product line manager. “This latest release continues to build on powerful VFX and finishing workflows in Flame with new features that support the latest HDR standards and integrate even more uses of A.I. to accelerate the creative process.”Useful for compositing, color grading, and cosmetic beauty work, the A.I.-based face segmentation tool automates all tracking and identifies and isolates facial features, including nose, eyes, mouth, laugh lines, and cheekbones, for further manipulation. Face matching algorithms are also capable of specialized tasks, including specific mole or scar isolation, through custom layout workflows. This builds upon Flame’s powerful A.I. feature set introduced in April 2019, with built-in machine learning analysis algorithms to isolate and modify common objects in moving footage, dramatically accelerating VFX and compositing workflows.To meet increasing demand for HDR content mastering driven by OTT streaming services, Flame 2021 introduces a powerful new Dolby Vision HDR authoring and display workflow. This enables Flame to import, author, display, and export Dolby Vision HDR shot-by-shot animatable metadata, streamlining creation and delivery of the most vibrant, vivid, high dynamic range imagery required by leading OTT streaming services. The update also expands collaboration with Autodesk Lustre and other Dolby certified color grading tools through enabling XML metadata import/export.New Flame Family 2021 features include:
- Machine Learning Human Face Segmentation: A new semantic keyer introduces the ability to isolate and extract alpha mattes for skin, eyes, lips, nose, and additional facial features via A.I. and machine learning with no manual tracking required.
- Dolby Vision HDR Authoring and Display Workflow: New HDR Dolby Vision implementation enables XML metadata import/export for collaborative workflows; adds Dolby analysis trim and target display viewing within Flame; and introduces eCMU, iCMU and HDMI metadata tunneling for viewing HDR content via SDR monitors and Dolby enabled TVs. Additionally, Flame’s multilayer edit timeline now includes a HDR track to keep Dolby Vision metadata in sync, make last-minute edit changes, finalize transitions, and integrate graphics.
- Physical Defocus: A new GPU-accelerated Defocus Effect makes it easier to add depth of field to shots and can be used in Batch as a standalone Matchbox node or in Image Selective in combination with a generated AOV Depth map. Choose between Physical and Action Camera focus models or create custom models by adjusting attributes. Lens artifacting and radial, chromatic aberration can be customized to achieve the level and quality of out-of-focus blur desired, Depth Slices can be increased and blended for more granularity of focus, and highlights can be bloomed using a selection of Bokeh attributes and lens noise models.
- Save and Recall Color Grading and VFX: Quickly save and recall color grading and VFX work in the new Explorer, a dedicated “grade bin” and reference comparison area to support artist workflows.
- Viewing Area: A new video preview mode shares artist viewports, including storyboard, manager, and schematic, to SDI or HDMI preview monitors. In broadcast mode, Gmasks can now be observed in the view area during editing along with any other tools you directly manipulate.
- GMask Pre-Made Shapes: New GMask pre-made shapes with softness are available for colorists, compositors, and finishing VFX artists in the image and action nodes.