Performance improvements via hardware accelerated video decoding/encoding.
Editing workflows (ripple editing, perhaps media management & three point editing).
Regular “maintenance”: small improvements, bug fixing etc.
Mentoring if we get GSoC projects or new community contributors.
We do not have estimates for how long each item might take; will figure it out as we go. In terms of staffing, only two people from BF work on sequencer (both part-time), so how much of the above will actually happen in 2026 is unknown. More community contributions would be most welcome!
Ability to use Image data blocks in VSE. Or perhaps, all Image strips should use Image data blocks, instead of referencing things by paths?
Ability for preview to show strips outside of main render area would be useful (for animating, layout, etc.). They would probably have to bypass modifiers/effects and be rendered directly in the preview area under regular “final image”.
Performance
Proxy workflows are still suboptimal: appropriate present proxy res should be picked automatically; proxies should never be stored on network shares etc.
GPU acceleration for media decoding
GPU acceleration for media encoding (there’s WIP PR from Sergey: #138080)
Preview area should operate on preview resolution
Scene strip performance issues:
Related to depsgraph
Related to preview using full “render” settings. Ideally preview would be “incremental” similar to how viewport is
Related to GPU resources being recreated each frame
“Fast path” when VSE is only used to cut/trim videos and output video settings match source format: full decoding/encoding could be skipped, and instead raw video/audio packets emitted to output
A handful of built-in audio effects (compressor, etc.)
Improve equalizer UI
Previous attempts: 117700 and 136363, kittcat on chat expressed interest.
Audio-to-subtitles via Whisper or similar
Could the models be downloaded on demand, maybe using the same machinery as remote assets?
“Push to talk”: press button, record input from mic into a sound strip. Possible issue: Audaspace does not seem to have microphone input functionality
It would be useful to have “master audio track” where global volume and global modifiers could be placed on.
(very long term) What should audio pipeline look like overall? If we want to eventually have ability to implement audio effects using geometry-nodes-like system, does it make sense to use Audaspace at all? Someone needs to figure it all out.
Does the plan make sense?
The above list is already way more than we will actually manage to do in 2026. But also, does the list make sense? Maybe something on the list is of way less importance than other items? Like and subscribelet us know!
And again, the amount of people working on VSE is very small. More contributions would be very welcome.
I tried color grading using ACES 2.0 config from ACES github.
I’ll be waiting for mask drawing on VSE.
For now. Let’s say…
The video clip is 10 minutes long.
Import the video to VSE.
Cut from 8 minutes to 9 minutes.
Import same video to
Movie Clip Editor to draw a mask.
Move timeline cursor to 8 minutes.
Draw a mask which is mostly use for rotoscoping.
Use that mask inside compositor.
Use it as Power Windows feature from DaVinci Resolve.
I’ll be redoing what video editor already done.
Masking videos in Movie Clip Editor at exact same in-out points,
for every single clip on the timeline.
If somehow, I can only draw mask directly which is already cut or
Sync video clip from Movie Clip Editor to exact cut point.
But then again, there could be many cuts from a single video clip.
Anyway, every improvement is good.
I’ll be waiting for exciting new features.
Indeed, masking on the Sequencer Preview would be extremely convenient. As a first step, for easing the process, maybe double-clicking LMB or just single RMB could lead you to open the source in the movie clip/image editor?
Please consider reworking these subpar quality and legacy operators (untouched on this side of 2.49):
Overlap Mode > Extend (it’s got multiple bug reports, and should basically be a full insert/ripple mode)
Move/Extend from Current Frame (It’s also got many bugs. Doesn’t display selection change. It executes from the Menu position, doesn’t respect Snapping etc.)
Frame Overlay (Unlikely to be used with the current UI, should be aligned with Industry Standard Comparison View features).
Metadata (also a legacy feature, not particularly user-friendly - one would expect to see metadata for all files here, but this is not the case).
If you’re going to add Image IDs to Image strips, you might as well add Text IDs to Text strips.
If you want more contributors, consider updating the list of good-first-issues. I’ve collected some over the years here:
Since Blender already ships with ffmpeg, why not integrate more of its features? FFmpeg can perform some very useful operations without re-encoding or “conversion”, like:
Changing Container Format (Remuxing).
This copies the existing video and audio streams into a new container format without touching the actual encoded data.
Extracting Audio or Video.
You can extract the audio or video stream from a file and save it as a separate file.
Trimming/Cutting Sections.
You can extract a specific segment of a video or audio file without re-encoding.
Merging/Concatenating Files (of the same codec and format):
For files with identical codecs and container formats, you can concatenate them without re-encoding.
Adding/Replacing Audio Tracks.
You can replace or add an audio track to a video file without re-encoding the video.
Those are some of the ffmpeg functions that would be very helpful to have in Blender.
Also, it would be nice if we could use ffmpeg commands inside of Blender.
It’s wonderful the list of possible / next improvements, event sound related.
I would call your attention to one point related to sound: functionality related to the flow of sound and possible effects at the end of the sound render. For example: we can have several sound strips in the VSE, and they “add”. A compressor at the end of the sound render will be essential.
I suppose that this is really hard, so, we could search for alternatives:
It could be possible to include the FFMPEG functionality, as an alternative renderer of the sound, and use its filters
or, perhaps, some kind of exporter to create a script that could help include the definition of audio strips in order to create the render in other product
I suppose that this is really hard, so, we could search for alternatives
My wild guess is that doing a “proper thing” is actually easier than doing any of your proposed alternatives. In specific part of Compressor, one issue is that Audaspace (the audio library that Blender uses) currently does not have a Compressor effect (there is a WIP PR but it has not landed yet). And then the “only” remaining issue, is that you’d need “some place” to define global sound settings or modifiers, i.e. like master volume or “master track” where the compressor effect could be put on. That right now does not exist, and someone would have to come up with UI/UX design for all of that, but on the “code side” is is not actually complicated… I think.
But! As things are right now, it is very likely that no one from BF will have any significant time to work on audio related things in 2026, so all of the sound related things would have to be community contributions.
the ui sounds like the easy part.
the master audio channel must live right there at the bottom, below the “channel 1”.
all the audio strips outputs must be routed to this master channel.
since channels are not selectable in blender, then we would need a new “master audio” strip. a type of strip meant to be used on the master audio channel. and thats the strip where the modifiers would be applied
an option to pin the channel would be nice of course.
about the location, thats where modern daws put it, and its great. most audio people are familiar with it.
I’m brainstorming ideas that make the most sense to me right now, I would love to make a new view of “channels” where it tracks channels that have items in them, and you can hide or pin specific ones, and it always shows the master, w volume + pan controls and output levels in db for all of them
I was thinking about this, I think it’s important to have both, because sometimes you do NOT want to pack your file inside the blend file (maybe the item needs to be updated at a later date, and you will load the new file in the same place)
unless an image datablock can already directly refer to a file and dynamically load it? then maybe it’s redundant to have it in the image editor AND the vse (or the options could be tied together through RNA)
