EWOCvj2 is a free GPL licensed open source modern "VJ" videomixer application aimed at live performance of videoprojection compositions, available for Windows, Linux and Mac.
Optional modules supply local (on your GPU) offline AI capabilities, consisting of:
POWER
• Unlimited number of layers.
• Unlimited number of effects per layer.
• About 300 effects, all GPU. Extend with ISF or FFGL.
• VIDVOX hap codec support(GPU decompression).
• Supports video resolutions up to maximum GPU texture resolution.
• Many video formats supported, with or without audio.
FLEXIBILITY
• Content drag and drop: Rearrange layers in stack. Move clips or layer clips (with control and effect settings bundled) between clip queue, layers and media bins. Drop from file explorer.
• Add mask stacks (analogous to layer stacks) to each layer or each effect. Recursively add masks of masks.
• Save, open, move to/from media bin: layer files, complete deck (A or B) files and complete mix files (both decks together).
• Scrapbook media bins allow visually arranging media on thumbnail grids in an unlimited number of named bins.
• Scenes allow four different layer stacks running together on deck level of which only one is actively displayed.
• Allows using either a preview system or direct live interaction with possibility to copy streams between the two modes and into the background scenes.
• Intricate mesh warping of layer images.
CONTROL
• Local crossfade between every layer with previous layers. .
• Dry/wet slider between effect and original.
• Wipes/transitions can be inserted locally (as a blend mode) as a composition aid (wipe position slider controlled) or as A to B wipe (crossfader controlled).
• 4 Main MIDI control decks can be set up to quickly change how main controls (speed, opacity, fast forward/rewind scratching, play/stop...) are changed on layer or deck level.
• MIDI learning and OSC for most controllable features.
• Loopstation for recording and looping parameter changes. Also allows parameter curve automation and FFT curve automation.
• Assign layers or mix monitor views to manual or AI assisted projection mapping outlines.
VERSATILITY
• Each layer has a clip queue.
• Supports live video inputs like webcams. On Linux every output can be sent as video4linux virtual webcam. On all operating systems NDI communication is supported.
• Easy on-monitor clip scaling and panning.
• Every output can be wired to an external display.
• Recording of main output to disk (HAP video).
• Media bin sharing for cooperation over network.
AI
• RECOnet neural style transfer training (allows creating realtime 'style' effects from inspiration images).
• Generative image AI (Flux 2): - text to image (you can reference up to ten reference images) - prompt driven edit image
• Generative video AI (LTX 2.5): - text to video - image to video - first frame last frame to video - first frames all frames lora: use edited first frame to to propogate edits to the rest of the video - character retention: use close-up photo to set actor in generated video - cutout guides: stationary cutouts moving over a video background get translated in realistic subject movement - enter a depth-pointcloud of your video inside a camera path editor that allows setting a new camera path for your video
• Segmentation (SAM3): cutting out objects from videos.
• Projection mapping (SAM3/MoGe): AI assisted creation of projection mapping outlines.
• Upscaling of images (REALESRGAN) and videos (EDVR and FlashVSR).
•
RECOnet neural style transfer training (allows creating 'style' effects from inspiration images)*
•
Generative image (Flux 2) and video AI (LTX 2.5)
•
Segmentation (SAM3): cutting out objects from videos
•
AI assisted projection mapping outline and depth detection (SAM3/MoGe)
•
Upscaling of images (REALESRGAN) and videos (EDVR and FlashVSR)
POWER
• Unlimited number of layers.
• Unlimited number of effects per layer.
• About 300 effects, all GPU. Extend with ISF or FFGL.
• VIDVOX hap codec support(GPU decompression).
• Supports video resolutions up to maximum GPU texture resolution.
• Many video formats supported, with or without audio.
FLEXIBILITY
• Content drag and drop: Rearrange layers in stack. Move clips or layer clips (with control and effect settings bundled) between clip queue, layers and media bins. Drop from file explorer.
• Add mask stacks (analogous to layer stacks) to each layer or each effect. Recursively add masks of masks.
• Save, open, move to/from media bin: layer files, complete deck (A or B) files and complete mix files (both decks together).
• Scrapbook media bins allow visually arranging media on thumbnail grids in an unlimited number of named bins.
• Scenes allow four different layer stacks running together on deck level of which only one is actively displayed.
• Allows using either a preview system or direct live interaction with possibility to copy streams between the two modes and into the background scenes.
• Intricate mesh warping of layer images.
CONTROL
• Local crossfade between every layer with previous layers. .
• Dry/wet slider between effect and original.
• Wipes/transitions can be inserted locally (as a blend mode) as a composition aid (wipe position slider controlled) or as A to B wipe (crossfader controlled).
• 4 Main MIDI control decks can be set up to quickly change how main controls (speed, opacity, fast forward/rewind scratching, play/stop...) are changed on layer or deck level.
• MIDI learning and OSC for most controllable features.
• Loopstation for recording and looping parameter changes. Also allows parameter curve automation and FFT curve automation.
• Assign layers or mix monitor views to manual or AI assisted projection mapping outlines.
• Make and run timecoded cue lists: MTC/LTC/ArtNet
VERSATILITY
• Each layer has a clip queue.
• Supports live video inputs like webcams. On Linux every output can be sent as video4linux virtual webcam. On all operating systems NDI communication is supported.
• Easy on-monitor clip scaling and panning.
• Every output can be wired to an external display.
• Recording of main output to disk (HAP video).
• Media bin sharing for cooperation over network.
AI
• RECOnet neural style transfer training (allows creating realtime 'style' effects from inspiration images).
• Generative image AI (Flux 2): - text to image (you can reference up to ten reference images) - prompt driven edit image
• Generative video AI (LTX 2.5): - text to video - image to video - first frame last frame to video - first frames all frames lora: use edited first frame to to propogate edits to the rest of the video - character retention: use close-up photo to set actor in generated video - cutout guides: stationary cutouts moving over a video background get translated in realistic subject movement - enter a depth-pointcloud of your video inside a camera path editor that allows setting a new camera path for your video
• Segmentation (SAM3): cutting out objects from videos.
• Projection mapping (SAM3/MoGe): AI assisted creation of projection mapping outlines.
• Upscaling of images (REALESRGAN) and videos (EDVR and FlashVSR).
