We’re excited to push an early development preview of new masking workflows to Premiere Pro (beta). Whether you are color grading, applying effects, or cutting objects out from the background, these new masking tools are built to help editors work faster, smarter, and more creatively without needing to go to After Effects for simple tasks.
Note: This beta is an early look at upcoming Premiere Pro workflows and is not intended to reflect general availability performance or polish. Please contact us with bugs, feedback, and feature requests to improve these features.
- Object Mask – Automatically identify objects and people in your footage using AI and isolate them with a single click. Plus mask refinement tools and fast tracking throughout clips.
- Redesigned Shape Masks – the traditional shape mask tools (ellipse, rectangle, pen) have been rebuilt from the ground up with many performance and UI improvements.
- Faster Shape Mask Tracking – the tracking algorithm has been completely rebuilt too. It’s significantly faster than before, more accurate, and offers more precise control for correction.
- Compositing tools – blend any combination of Object and Shape Masks to create complex masks with new blend modes: add, subtract, intersect
To get started
New masking workflows are only functional in beta projects that support an upcoming major release of Premiere Pro. This means your beta projects will not open in Premiere Pro 25 or earlier. We're excited to give you early access with these steps.
- Update to Premiere Beta version 25.6 build 48 or later.”
- If you see this dialog, choose “Enable new masking”
Launch dialog
– OR – if you do not see the dialog at launch, or you skipped it, do the following… - Open or create any project
- Enable “New Masking” via the beta menu in the OS menu bar at the top of the screen. Premiere Pro (Beta) will need to relaunch.
Beta menu
- From now on, Premiere Pro (Beta) will create projects in version 26.0 format.
WARNING: these projects are NOT compatible with the current shipping version of Premiere Pro and will not open.
Quick Start guide
Follow these steps to get started with New Masking.
- Make sure you have enabled New Masking (see above).
- Select a clip on the timeline.
- Find the new Mask Tool group in the toolbar. Long-press to see all the masking options. Choose the Object Mask tool.
Tools
- Hover the mouse over the program monitor, watch as objects are identified, and click. This will draw a static mask around the object or person.
- Track the motion. Find the new “Object Mask” in the “Unassigned Masks” section of the Effect Controls Panel. Click on the forward and back button.
Tip: the Effects panel is a different panel from the Effect Controls Panel. Find it in the Window menu.
- Add an effect. Find the Gaussian blur effect in the Effect Panel. Drop it on the clip in the timeline. The mask will automatically be applied to the effect.
- Congratulations, you just used Object Mask to blur an object!
- Now, try to use the other tools in the group in a similar manner.
Visit here for the complete documentation. https://helpx.adobe.com/premiere-pro/using/object-masking.html
We want to know what you think. Join the conversation below.
2 respostas corretas
@FlyingFourFun Thanks for your detailed feedback.
I will reach out to you over direct message regarding how to share the videos you were referring to.
We have a helpx page that might help with some of the questions you have raised.
https://helpx.adobe.com/premiere-pro/using/object-masking.html
Blend mode of mask can be used to combine the masks together. You can use "Add", "Subtract" or "Intersect" to combine masks. Any number of Object Masks and Vector Masks can be combined to create complex mas
Make sure to right click on the tracked mask in the Effect Controls and select Change to Opacity Mask. Duplicate your video and remove the tracked opacity mask and placed that on V1 and your masked layer on V3 or above. Place your graphics on V2 and it should show up.
135
respostas
Finally a moment to do some testing with the new mask features. Here's my feedback after a quick spin.
As overall feature this is a great step forward. Allowing tools to be expanded in functionality by adding UI elements in the viewer (program monitor) is a good choice. Many other tools can potentially benefit from such mechanisms breaking out of the sometimes restrictive effects panel slider, checkbox, menu principles. Why not promote the tracker buttons to the Program Monitor as well? So you can keep your eyes around the same working area when tracking.
- Individual mask tools cannot be cycled with a single assigned keystroke (like in Photoshop or AE) making cycling between them a pain. Assigning a key per tool is not a solution.
- In addition to the above, what also helps is if the mask tool types are exposed in the program monitor UI so you can switch between them quickly after either of them was activated, without using keyboard shortcuts or holding down your mouse while selecting it in the toolbar. See this mockup for reference + the tracker tools.
- When transforming, anchor point cannot be snapped to one of the 9 boundingbox positions. Would expect this to be possible via holding CTRL while moving it for example.
- Transform controls for rect/elipse is confusing w.r.t their values. Anchor becomes 0,0 (top left canvas), size is the sequence resolution, shape returns to original drawn position. Why not make this function the same as AE's shapes where rect path size == absolute in pixels drawn and transform size(scale) = separate % scalar with 100% == drawn size. Such a mechanism would also allow for drawing fixed resolution masks onto footage.
- Would be useful if center positional data of a tracked mask could be used to drive other effects. For example you track a face and you copy paste it to transform of a text element. (See how BorisFX Mocha stores tracking data)
- AI masking option is rough. There is severe lack of temporal stability or mask detail around edges and transparency. There are also no options to refine the matte with post effect other than feathering/blur. To make it more useful you'd need some user controlled tools like painting in gaps created on random frames by the model.
- Editors likely use masking tools 99/100 times for opacity. This should be a default (or user preference) if no other destination is selected when drawing on a selected clip, instead of the unnassigned masks section.
To conclude, what definitely a showstopper is atm is the fact that it takes way too much time getting from A to B if all you want to do is apply an elipse or rectangle mask onto a clip. If point 1, 2 and 7 are addressed this should already be greatly improved.
No problem @Ron Rigler ! The point isn't so much how to draw shapes, but the way data is handled is not efficient and artist friendly.
What currently happens is that the property [Scale] is upon first draw declared the dimensions of the sequence regardless of the drawn shape.
This in itself is already unusable because it doesn't reflect the ratio of the shape.
When manipulating the small shape to roughly match the larger shape, you get dimensions that are totally meaningless to the user from an information parsing point, and due to slider sensitivity roughly based around values up to say 1000, make sliding them instead of changing the shape in the viewer totally useless.
This is why it would make much much more sense if dimensions for Elipse and Rectangle were absolute pixel based.
Let's promote a new value [Size] (being a property of the path/object). In addition there could be a secondary (existing) [Scale] slider (part of transform) that at 100% (default value) reflects the shape's pixel size but can be scaled as secondary operation. This mechanism would allow animating such a shape from 0-100% but using pixel width/height to determine the real dimensions for it. This would tremendously help artists type in say 300x600 and (provided media clip matches sequence res and isn't by itself scaled) create a mask that is exactly that rather than get some arbitrary values from current mechanism.
Here's a mockup of what I mean. This way scale is also able to be reset to 100% rather than back to sequence size. The initial draw of the shape dictates [Size] in pixels instead, and so should scaling it via the path in the viewer.
Even better would also be the ability to change between Rectangle and Elipse on the mask itself. Since these are 'fixed' shapes they can easily be made dynamic and get a dropdown that allows toggling between them. When adjusting a point of a rectangle that makes it not a rectangle anymore, the Size property should reflect the dimensions of the bounding box instead and keep relative shape when altered.
When flipping back and forth between Elipse and Rectangle the previously used shape data should restore itself.
Lastly, once any point is altered in a Rectangle there seems to be no way to recover the real rectangle (90degree angles). There should ideally be an option for this for cases where you did so accidentally and can't undo anymore.
These ideas are based on how AE shape layers and the new Mesh primitives in v26 are designed. I think it's much more rugged this way and it would also help a little bit with mechanism parity between these software packages.
AdChoices