Looks ready
Microsoft Windows · Share · Photo edit & compress
Designing the Last Mile of Sharing in Windows
The photo was ready to send. The correction arrived one second too late.
I redesigned Windows Share so people could correct and resize a photo at the moment they noticed the problem, without losing the destination or restarting the journey.
My roleI led the interaction design for the editing and compression journey, from mapping the broken handoff and defining the canvas model to prototyping the tool hierarchy, file-size choices and transition between compact sharing and focused editing.
One continuous flow Edit + compress Adaptive Share Canvas System-level Windows pattern Covered in global press
01 · The moment of commitment
People often notice the problem only when the photo is beside its destination.
An image can look perfectly acceptable while it is sitting inside a folder.
The moment it is placed beside a person, application or post, the standard changes.
A distracting edge becomes visible. The subject feels too small. The photo is sideways. A private detail needs to be removed. A file-size limit suddenly matters.
These are not always planned editing sessions. They are small corrections triggered by the act of sending.
That made the Share Sheet an important moment in the journey. It was the first place where the image, destination and consequence became visible together.
But Windows treated that moment as too late to make a change.
One photograph · two contextsThe destination changed the judgement
Now the problem is visible
Distracting edgeDetail to remove4.8 MBThe destination changed how the person judged the image.
02 · The cost of leaving
A ten-second correction could become a complete restart.
Once someone noticed the problem, the existing recovery path moved them away from everything they had already decided.
They had to close the Share Sheet, open another application, locate the same image, make the change, save a new version and return to sharing.
The original destination was no longer part of the experience.
The person had solved the image problem by losing the sharing context.
The steps were individually understandable. Together, they turned one small correction into a fragmented journey.
The existing path · 13 stepsThree decisions repeated
01Choose photoIntent begins
02Open ShareFirst decision
03Choose destinationFirst decision
04Notice problemCorrection appears
05Close ShareContext disappears
06Open editorNew task begins
07Find photoFirst repeat
08EditThe actual correction
09Save copyAnother file appears
10Find edited copyPhoto repeated
11Open Share againShare repeated
12Reselect destinationDestination repeated
13SendOriginal goal
Do not make someone restart after they have already decided to send.
03 · The reframe
This was not a creative editing session. It was preparation for a destination.
The first framing treated the feature as a smaller photo editor inside the Share Sheet.
That immediately created the wrong questions. How many tools should it contain? How closely should it resemble Photos? Which editing features could fit into the panel?
The better question was: what does someone need to correct before this particular image can leave?
The experience did not need to reproduce a full creative application. It needed to support the small, high-confidence actions that frequently appeared at the point of sending.
Crop the image. Correct the orientation. Mark something important. Remove something distracting. Adjust the visual enough to make it usable. Choose an appropriate balance between quality and file size. Then continue sharing.
Last-mile preparationOnly the actions needed before sending
04 · The constraint
The Share Sheet was compact by design. Accurate editing needed room.
The central design problem was spatial.
A sharing panel should stay lightweight. It appears temporarily, helps someone choose a destination and gets out of the way.
An editing canvas has different requirements. The person needs enough image area to judge the crop, compare adjustments, manipulate handles and understand whether quality has been lost.
Trying to preserve the original panel width made the Share Sheet look consistent. It also made the editing experience too small to use confidently.
The surface could remain compact. Or the editing task could remain useful. It could not do both at the same time.
One image · two geometriesBehaviour over dimensions
Consistent footprint, unusable judgement
- The image becomes too small
- Crop handles crowd the preview
- Controls compete with content
- Quality changes are hard to judge
Different footprint, trustworthy decisions
- The image remains legible
- Manipulation targets have space
- Tools separate from content
- Before-and-after becomes inspectable
I chose to let the system surface change shape rather than make the editing task conform to the wrong shape.
05 · Exploration
How much space is enough to make a trustworthy decision?
Making the panel larger was not a complete answer.
The surface still needed to feel connected to sharing rather than opening an unrelated application inside the flow.
I explored three levels of expansion: editing inside the compact panel, a full-screen editor, and an adaptive Share Canvas.
I chose the adaptive canvas. It gave the image priority without turning sharing into a full creative workspace.
Three spatial directionsSame image, destination and controls
Editing inside the compact panel
Preserved the Share Sheet footprint but reduced the photo to a preview too small for accurate manipulation.
Full-screen editor
Created the best editing space but visually separated the image from the destination and made sharing feel like a new task.
Adaptive Share Canvas
Expanded enough for meaningful editing while keeping the file, destination and send action connected.
06 · The system
The surface expanded with the task, and contracted with completion.
The final model treated the Share Sheet as an adaptive surface.
In its resting state, it remained compact and destination-focused.
When someone selected Edit, the surface expanded around the image. The destination did not disappear; it moved into a quieter supporting role while the editing task became primary.
After the change was applied, the surface returned to the compact sharing state with the original destination and context intact.
The experience should never feel as though Windows launched another application and later brought the user back. It should feel like one surface briefly creating enough space for the current decision.
PreviewFile and destination visible
Enter editImage becomes primary
ExpandCanvas creates room
ApplyOutput preview updates
ReturnContinue to same destination
The surface changed state. The journey did not change direction.
Expanded Photo Edit stateThe spatial decision in the project artifact
After ApplyThe corrected file, ready to send
07 · The flow
Preview. Correct. Size. Send.
I organised the interaction around four decisions.
The person first inspected what they were about to send. They corrected the image only when necessary. They chose the appropriate file-size trade-off. Then they continued to the destination they had already selected.
The interface avoided asking someone to understand every editing and compression option at once. Each decision appeared when it became relevant.
Preview → Edit → Apply → Compress → SendDestination persists in every frame
Outlook selectedInspect the image, file name, size and destination.
Outlook preservedExpose the focused correction tools only when requested.
Outlook preservedReturn to a clear output state with undo available.
Outlook preservedChoose a consequence-led size and quality trade-off.
Outlook selectedComplete the original task without selecting the destination again.
08 · Separating intent
Editing changes the image. Compression changes what the image can carry.
Editing and compression appeared at the same moment, but they solved different problems.
Cropping, rotating and adjusting changed the content or presentation of the photo.
Compression changed the trade-off between visual quality and the cost of sending it.
Combining them inside one undifferentiated toolbar would make that distinction difficult to understand.
I kept both capabilities close to the image while giving them separate moments in the flow.
Two connected decisionsContent first, delivery second
Does this photo look right?
Prepare the image.
ContentCompositionOrientationClarity
What is the right version to send here?
Prepare its delivery.
Visual qualityFile sizeDelivery speedDestination limits
09 · Making technical trade-offs human
Nobody should need to understand a codec to send a smaller photo.
Compression interfaces often expose technical values without helping people understand the result.
A percentage or quality number can appear precise while communicating very little.
I designed the choice around consequences rather than implementation.
Sharp preserved the most visual detail. Balanced reduced the file meaningfully while keeping it suitable for most sharing contexts. Small prioritised a compact file for constrained destinations or slower connections.
Each option showed the estimated output size before commitment, while the preview remained available so the choice did not depend on a label alone.
Compression choicesShow the consequence, not the codec
Original 4.8 MB
Original remains unchangedChanges apply only when you send or save a copy.
Original shipped screenCompression inside the Share editor
Show the consequence, not the compression vocabulary.
10 · Tool hierarchy
More capability would have made the moment less useful.
Once the canvas expanded, it became tempting to add more editing tools.
The larger surface could technically accommodate filters, detailed colour adjustments, cleanup tools and advanced markup.
But the product was still the Share experience.
Every additional control increased the chance that a quick correction would turn into an editing session, and that the destination would disappear from the person’s attention.
I used progressive disclosure to protect the primary journey. The first level focused on the most common last-mile corrections. Secondary tools remained available without competing with crop, orientation and compression.
The surface was allowed to grow. The scope was not.
Grounded in the toolbarA clear boundary around the Share experience
Primary
CropRotateMark upCompress
Secondary
AdjustFilterClean
Outside Share
Advanced retouchingLayer editingComplex colour gradingAsset management
11 · Precision
A quick edit still needed to be reversible, inspectable and accessible.
A lightweight experience could not become a careless one.
Cropping and compression alter what leaves the device. People needed enough control to understand the output and recover from a mistake.
I designed the flow to preserve the original, keep undo visible, support before-and-after inspection, show file-size consequences before commitment, support keyboard and touch, and respect reduced motion.
The editing controls, crop handles and tool switching needed predictable focus, sufficiently large targets and alternatives to precise pointer-only manipulation.
Craft review viewSafeguards around a lightweight edit
UndoKeep the last correction reversible.
Original preservedThe source image is not silently replaced.
Output sizeShow the consequence before commitment.
Focus and targetsKeyboard-visible, touch-sized controls.
12 · System scale
The design could begin in one Share Sheet and matter across Windows.
The built-in Windows share window is not tied to one application.
Microsoft documents the system surface across the desktop, File Explorer, Photos, Snipping Tool, Xbox and other applications that invoke the built-in Share window.
That made the canvas decision larger than one image editor.
The interaction needed to work as a system pattern: recognise when the shared object can be prepared, expose an appropriate action, expand only when the task requires it, preserve the source and destination context, and return cleanly to sharing.
The exact tools could change with the file type. The behavioural contract could remain consistent.
System-design scopeOne adaptive contract, many origins
Invoked from
DesktopFile ExplorerPhotosSnipping ToolXboxOther Windows apps
13 · Outcome
A destination picker became a place to finish the image.
The project reframed photo editing and compression as part of the last mile of sharing.
Instead of forcing people to leave after discovering a problem, the adaptive canvas gave the image enough space for meaningful correction while preserving the destination that made the correction necessary.
The work connected two previously fragmented activities:
Prepare the image. Send the image.
The design established a clear interaction model for Windows Share: remain compact while choosing a destination, expand when preparation requires visual space, keep editing and compression understandable, preserve the original, return to the same sharing context and complete the journey without starting again.
The value was not the number of tools inside the panel. It was removing the boundary between noticing a problem and being able to fix it.
1 continuous journey
Edit, compress and send without rebuilding context
The destination survives the correction, so the person never has to reconstruct why the image was open.
2 connected decisions
Prepare the image and prepare its delivery
Editing and compression stay close while answering different questions at distinct moments.
1 adaptive system surface
Compact for sharing, expansive for precision
The surface changes geometry without changing the direction of the journey.
Across Windows Share
A pattern for a system surface invoked from multiple applications
The file type may change. The contract to preserve source, destination and intent can remain consistent.
-
01Share openedFile and destinations remain visible. -
02Edit in contextThe canvas expands inside Share.
-
03Apply changesThe updated image commits in place.
-
04Continue sharingReturn to the same destinations.
14 · Reflection
Sometimes the right solution is not another application. It is enough room at the right moment.
This project began as a request to add editing and compression to the Share Sheet.
The deeper design problem was deciding whether the surface itself could change.
Keeping the panel compact would have preserved visual consistency. It would also have forced people to make image decisions on a canvas too small to trust.
Expanding to a full editor would have solved the spatial problem while breaking the sharing journey.
The adaptive canvas sat between those extremes.
It respected the temporary nature of the Share Sheet while giving the image enough priority when precision mattered.
A system surface does not need to remain the same size to remain the same product. It needs to preserve the person’s intention as the task changes.
Keep the journey compact. Give the decision room.
Confidentiality note
This project was completed at Microsoft. The case study uses public-safe, reconstructed or approved product material. Internal research, implementation details and roadmap information remain confidential.