Contextual Task Bar - Generative painting for watercolour and oil painting effects with quick mask
Hello Adobe Photoshop Team,
I would like to report a regression affecting the Quick Mask and Generative Fill workflow used to create watercolour, oil-painting and other generative art effects.
Adobe previously demonstrated this workflow using Quick Mask to create an intensity-based selection, followed by Generative Fill through the Contextual Task Bar:
https://www.youtube.com/watch?v=4THG0IPx9hc
https://www.adobe.com/learn/photoshop/web/create-art-styles-generative-fill
Workflow that previously worked
- Open an image in Photoshop.
- Enter Quick Mask Mode.
- Fill the mask with a controlled percentage of grey to determine the strength of the generative transformation.
- Exit Quick Mask Mode to convert the mask into a selection.
- Use Generative Fill from the Contextual Task Bar.
- Enter a prompt such as “watercolour painting” or “oil painting.”
This produced highly useful painterly interpretations while retaining the composition and visual structure of the original image.
Current problem
In newer Photoshop updates, the Generative Fill option no longer functions for this art effects after creating the selection through Quick Mask Mode.
The technique worked successfully in earlier versions but has been unavailable or inconsistent for approximately two years.
This appears to be a regression rather than simply a change in interface, because the Quick Mask method allowed something that ordinary selection tools do not: precise control over the intensity of the generative effect through tonal mask values.
Expected behaviour
After exiting Quick Mask Mode, Photoshop should recognise the resulting selection and make Generative Fill available through:
- the Contextual Task Bar;
- the Edit menu; and
- any relevant Generative Fill panel.
Selections created through Quick Mask should behave consistently with selections created using the Marquee, Lasso, Object Selection and other selection tools.
Why this matters
This was an unusually creative and valuable workflow for artists, illustrators, designers, photographers and educators. It allowed users to combine their own image-making decisions with generative tools rather than replacing the entire image with an uncontrolled AI interpretation.
The adjustable grey value acted almost like a creative 'strength' control, allowing the user to decide how closely the generated result followed the source image.
I use Photoshop professionally and teach creative software to art and design students. This technique was particularly useful for demonstrating:
- tonal masks and selection logic
- non-destructive experimentation
- the relationship between source imagery and generative interpretation
- watercolour, oil, drawing and print-inspired visual treatments, and
- responsible use of generative AI within an artist-led workflow
Requested fix
Please restore full compatibility between Quick Mask selections, the Contextual Task Bar and Generative Fill in both Photoshop and Photoshop Beta.
It would also be helpful for Adobe to test this specific workflow across different mask densities—for example 20%, 40%, 60% and 80% grey—to ensure that tonal intensity continues to influence the generated result as expected.
Feature request: Generative Painting
Please also consider developing this workflow into a supported feature, perhaps called Generative Painting or Generative Style Transfer.
A dedicated feature could include:
- a strength or source-image preservation slider
- watercolour, oil, gouache, ink, pen and ink, pencil drawing, charcoal drawing, scribble drawing, pastel, and printmaking presets
- control over brush size, pigment texture and level of abstraction
- the ability to use a greyscale mask to vary the effect locally
- reference-image support
- preservation of important edges and composition
- generation onto a separate, editable layer, and
- the option to combine painted and photographic areas within one image
This would turn an inventive workaround into a properly supported artist-facing tool. Photoshop is most powerful when AI extends the artist’s hand rather than removing it from the picture.
Thank you for investigating this regression and considering the feature request.
Kind regards,
Kavitha Paramasivan
Graphic Designer, Artist, and Creative Software Educator
