What happens:

1. Take a photo/video or select a photo to add
2. Drag the thumbnail to delete it
3. Kanvas closes and you return the main WP view

What I expected:

1. Take a photo/video or select a photo to add
2. Drag the thumbnail to delete it
3. Return to the camera capture view to take a new photo or select a new photo

Allow Kanvas to support the landscape device-orientation, so landscape media would fit better on the screen, and the camera would shoot landscape-oriented media.

Also, if we can adjust to a dynamically-changing screen size, this would also allow us to support the iPad.

The button has to travel quite a distance and is rather distracting. I think a simple cross fade here would be fine instead.

Kanvas should be usable with Swift Package Manager.

Not totally sure if this is an issue with Kanvas or WordPress iOS but the video file from Kanvas is uploaded directly so I'm guessing it's a format problem.

We should add the following logging methods to KanvasCameraAnalyticsProvider and call them at the appropriate times:

• logCameraPermissionView
• logMicrophoneAccessInitiated
• logCameraAccessInitiated
• logMediaPickerOpen
• logMicrophoneAccess(notDetermined|restricted|denied|authorized)
• logCameraAccess(notDetermined|restricted|denied|authorized)
• logMediaPickerAccess(notDetermined|limited|restricted|denied|authorized)

Support GIF stickers.

For photos, the photo will be converted to a video of a set duration, and the sticker will loop over that.

For GIFs or videos, the sticker will just loop over the video.

• Accept permissions when first using
• Go back to the Privacy screen in settings and disable camera and microphone access
• Go back to create with Kanvas
• The "Allow access to camera/microphone" buttons do not work

We probably need to show better messaging here as I don't think it's possible to enable the in-app after they have been manually disabled.

The MediaPlayerTests.testPlayVideo() method fails run run on CircleCI:

Test Case '-[KanvasCameraExampleTests.MediaPlayerTests testPlayVideo]' started.
2020-10-27 01:10:30.268425+0000 KanvasCameraExample[1467:68019] [aqme] AQMEIO.cpp:352:AQMEIOManager_FindIOUnit: error -66680 finding/initializing Device.AQDefaultDevice
2020-10-27 01:10:30.268629+0000 KanvasCameraExample[1467:68019] [aqme] AQMEIO.cpp:352:AQMEIOManager_FindIOUnit: error -66680 finding/initializing Device.AQDefaultDevice
2020-10-27 01:10:30.268791+0000 KanvasCameraExample[1467:68019] [aqme] AQMEIO.cpp:352:AQMEIOManager_FindIOUnit: error -66680 finding/initializing Device.AQDefaultDevice
2020-10-27 01:10:30.268965+0000 KanvasCameraExample[1467:68019] [aqme] AQMEIO.cpp:352:AQMEIOManager_FindIOUnit: error -66680 finding/initializing Device.AQDefaultDevice
2020-10-27 01:10:30.269141+0000 KanvasCameraExample[1467:68019] [aqme] AQMEIO.cpp:352:AQMEIOManager_FindIOUnit: error -66680 finding/initializing Device.AQDefaultDevice
2020-10-27 01:10:30.269486+0000 KanvasCameraExample[1467:68019]  AudioObjectRemovePropertyListener: no object with given ID 0
2020-10-27 01:10:30.269625+0000 KanvasCameraExample[1467:68019]  AudioObjectRemovePropertyListener: no object with given ID 0
2020-10-27 01:10:30.269743+0000 KanvasCameraExample[1467:68019]  AudioObjectRemovePropertyListener: no object with given ID 0

Restarting after unexpected exit, crash, or test timeout in MediaPlayerTests/testPlayVideo(); summary will include totals from previous launches.

Add shields to the Readme:

• Cocoapods
• CircleCI

Currently, Stitch mode will store captured media on disk, but the thumbnail is stored in memory. This means we could run out of memory if someone captured a large number of images or videos.

The thumbnails should be on disk also, and we should make sure that they are loaded out of memory as the corresponding cell in the clips list goes off-screen.

Right now we block the camera from being used until you grant microphone and camera access.

Instead, we could just require one of them, and respond accordingly:

• Just microphone access would show an audio-only recorder
• Just camera access would still record video, just without audio

In both cases, we'd have to show that we're missing the other permission, and provide a way to navigate to System Settings to enable it.

The toolbar icons can be difficult to see on some photos with lighter colors.

Perhaps a gradient could be added behind to make them stand out more?

pbMoDN-1Hv-p2#comment-3573

We should use the newer iOSSnapshotTestCase for several reasons:

• More actively developed than FBSnapshotTestCase
• Includes support for XCTAttachments (should make failures easier to parse)

Tapping on the text field should bring up the copy and paste menu.

You can technically copy by using the 3 finger pinch-out gesture, but this is not intuitive.

Unless we can easily use different filter's that feel at home in the WP apps.

This test failed routinely when running in Tumblr's CI, but not sure about CircleCI. Regardless, we should make sure this test is sound and enable it.

We'll want to add an sound on / off toggle button on video slides so the user can turn audio off for a video.

The FilteredInputViewControllerTests.testFilteredInputView fails to run on CircleCI:

Test Case '-[KanvasCameraExampleTests.FilteredInputViewControllerTests testFilteredInputView]' started.
Fatal error: Failed to create MetalContext: file KanvasCamera/FilteredInputViewController.swift, line 25
2020-10-27 01:10:02.339476+0000 KanvasCameraExample[1452:62258] Fatal error: Failed to create MetalContext: file KanvasCamera/FilteredInputViewController.swift, line 25

Restarting after unexpected exit, crash, or test timeout in FilteredInputViewControllerTests/testFilteredInputView(); summary will include totals from previous launches.

Currently all media is exported as its original resolution. While that works great most of the time, someone might want to zoom in on a landscape photo and just use a portion.

Let's add a screen-aspect-ratio mode which allows you to position the media inside the screen as you see fit, and choose a background style (color/gradient/blur).

We should review the Permissions screen and ensure it fits in with other patterns in WordPress iOS.

iphone 12, running iOS 14

This is hard to describe so I made a video:
https://cloudup.com/crMQVKpdxlu

Unless you're drawing in the center of the screen the marker is applied to the left or right of your finger. This makes it really hard to write with the tool or draw a symbol or shape. As you're writing or drawing across the screen the position changes and it's unpredictable.

A text resizing control should be added while editing text. There is no way to avoid multi-line text otherwise.

A user can create a sticker by tapping a button in the sticker drawer, which will open a new instance of the Camera, allowing them to either capture media or pick it with the media player, and send them to the Editor. However, instead of posting, the next screen will let them cut out the sticker, either with a few preset shapes (circle, square, quote) or freeform. The user can then confirm the sticker, and it's added to their sticker drawer. This will only be persisted in the app.

Icon-Story.svg.zip

Can't really see the case for this functionality in stories. It might be interesting to keep the video trimming functionality for a future version, but not v1.

"Open Kanvas Dashboard" is meant to mimic a swipe-to-open interface, which would be useful to have, but this one crashes on the 2nd open, and the code it's based on is now significantly different in Orangina, so might as well just make one from scratch here, or totally remove it.

For the WP Stories instance, let's swap the "Post to Tumblr" text for:

Allow camera and microphone access

Right now Stitch mode just let's you use the camera.

Let's enable the media picker for Stitch mode. The final media should be the resolution of the first image, and the remaining images should be aspect-fit inside that, with the negative space being the image aspect-filled, but blurred.

Currently, if someone denies Photo Library access, we'll just don't save media to their Photo Library, but we will still call didCreateMedia to allow the host to use the media.

It'd be nice in the Editor to somehow indicate (in a non-blocking way) that they've denied this permission, and link them to Settings to enable it if they choose.

I expected to have access to the predictive text options that are usually displayed above the keyboard. (I've come to rely on them too much.)

Also noticed they aren't displayed above the keyboard in the publish sheet. I'd expect to have them there too.

Tried drawing on a story and it looked like nothing was happening or showing up in the editor. When the story was exported the drawing was visible.

Here's a video of the editor:
https://cloudup.com/cuihC7XqDMq

And this was the output:
https://wpstoriesdemo.mystagingwebsite.com/2020/11/10/palm-tree/

Testing 16.3 (39815)

Videos in this story are rotated 90 degrees counter clockwise. They're both portrait videos that appeared right in the editing view.

The pod should be named Kanvas and the framework should be called Kanvas. This project is not just a camera, so let's have the name reflect that.

Add the ability to delete frames using the Trash view by dragging and dropping frames.

Where do the stickers in Tumblr come from? Is it a custom library? Or sourced from a third party?

Could we easily reuse the UI here to show the device emoji library?

Noticed in the 35257 build that you can't toggle to change the font between the sans serif and the script. This button works in the regular tumblr build.

Normal text fields in iOS allow you to move the cursor by tapping a new location. The text field in Kanvas appears to swallow those touch events so that the cursor stays in place. It can still be moved by dragging or tapping and holding.

Current behavior:
Both landscape and portrait media is scaled to fit the width of the screen.

Options for discussion:

1. Automatically scale the media to fit the full height of the screen for portrait images. Continue scaling to fit the width for landscape.
2. Allow for manipulation of the background image so user can scale and crop.
3. Something else?

Right now, if Stitch mode allows using media with different resolutions, there's some problems:

1. The preview skips clip that's are a different resolution than the first clip.
2. The output stretches all clips to be the size of the first clip.

Ideally we'd have the output media be the resolution of the first media, and subsequent media gets centered and aspect-filled.

On shorter height devices, the controls may be covered by the multiple story bar.

1. Add text to a slide, leave it in it's original spot
2. Add more text to the slide - the text from step 1 is visible and it's hard to see what you're doing in the new text edit mode

Right now, you have to drag each clip you've captured in Stitch mode to the trash to switch to another camera mode.

There should be a way to switch modes without needing to delete each clip individually. Either a way to delete all the clips quickly, or, preserve stitch clips.

Currently the GIF maker converts the entire video to a GIF, even though we only export max-3 seconds. While this works for shorter videos, long videos can take very long to convert, and we have to do it at a poor quality.

Instead, the GIF maker should work against the video itself for seeking and such, but in the background convert it to a GIF at a good quality. When the conversion is complete, it should swap the media out so it plays back the GIF.

Playback modes like rebound/reverse won't work against the video, but that's OK.

When swapping the video and GIF, playback should be seamless, and not restart.

(Having this setup this way will allow us to just use the trimmer for videos too!)

When enabling Metal for the rendering pipeline, the camera viewport is not centered:

OpenGL	Metal

This is most likely due to the camera viewport cropping that happens in OpenGL to ensure camera-captured media has the same aspect ratio as the screen.

The Revert button in the GIF Maker should be localized:

Repro steps:

1. Switch to GIF mode
2. Capture a GIF by tapping the shutter - this will capture a 1.5 second GIF, and open the GIF Maker
3. Tap and hold on the end-trim handle

Expected: it should show 0:01
Actual: it shows 0:00