Repository navigation
Conversation
e8edd92 to
0004bf2
Compare
0004bf2 to
a13cbc7
Compare
|
Please re-review. The double duration (intended to mimic the config) is resolved to just one time. And the shader path now is it's raw data instead. I updated the code with force push to keep 1 commit and keep things clean. Let me know if you like that or if you rather have individual commits from now on. To me either way works so just let me know your preference here. |
|
@codex review please |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a13cbc762b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
635af4f to
bdb4c04
Compare
|
@codex review please |
|
To use Codex here, create a Codex account and connect to github. |
|
Ahh, that doesn't work if i do it :) So i spend a good day getting this in a working state again. Not just the changes here but also the companion changes. This should be good to go now. Can you re-review? |
|
It's becoming hard to keep this working, please review! I want to merge this for the next release if possible. @vaxerski |
|
I can't review this when it has conflicts |
Yeah, this sings a little. There were no conflicts 2 weeks ago. They did start 2 weeks ago around the same time. Which i completely missed. Please, review within a day or so after i fix this one (I'll let you know when it's fixed) as it's annoying to resolve merge conflicts when there were none and i think this one has been here long enough by now. Any particular reason for the two companion PRs to not be merged yet? That's hyprwm/Hyprland#15347 and hyprwm/hyprpaper#371 nether has merge conflicts at the moment. |
|
because they rely on this API? This needs to be merged + in a release first |
Depends, they rely on a message format change that is introduced in this one: hyprwm/Hyprland#15347 which i think needs to land first. Nothing breaks with it, but it prepares the message format for the change. Obviously only if you approve of the idea conceptually. The sensible way to me is:
|
|
there is no point in landing the message format before this. Adding a feature that doesn't work is pointless IMO |
bdb4c04 to
9319696
Compare
…aders Adds an animated wallpaper transition path to CImageElement: a two-texture blend (start -> end) driven by fadeLayersIn/Out durations, with an optional custom GLSL fragment shader supplied by path. - CImageElement::transitionTo captures the in-flight blend to an FBO when a transition is interrupted mid-way, using that as the new start texture so a fresh transition can begin to the new target (no flicker, no black flash). - IRenderer gains renderTransition / captureTransitionState / ensureTransitionShader. Custom shaders are stat()-checked once at transition start and recompiled only on mtime change; the per-frame render path is a single cache lookup + draw, with no filesystem or compile work. - CFramebuffer sets m_tex->m_size after glTexImage2D so captured textures report the right size to the shader's tex1Size/tex2Size uniforms. - A captured texture keeps its owning CFramebuffer alive via m_tempFramebuffer. - All fit modes (stretch/cover/contain/tile) are honoured; the host pushes the config duration to u_duration each frame (GLES cannot initialize uniforms).
9319696 to
bf7e306
Compare
|
@vaxerski Enjoy the review :) Rebased against current master. If you want to try it out then you need the two companion PRs (also rebased where needed). Now say you want to have a circle reveal you can use this shader: Do note that this PR intentionally contains a The following command would show it all if you applied the PRs and use the above shader. |
vaxerski
left a comment
There was a problem hiding this comment.
also format your code with clang-format please. Those comment-related comments arent exclusive, at least half the comments here are redundant
| bool isTransitioning() const; | ||
|
|
||
| float getTransitionProgressOut() const; | ||
| float getTransitionProgressIn() const; |
| uniform sampler2D tex1; // Start frame (fit pre-resolved in C++ via renderFitFrame) | ||
| uniform sampler2D tex2; // End frame (fit pre-resolved in C++) | ||
| uniform float progress; // 0.0 = fully start, 1.0 = fully end | ||
| uniform float alpha; // Overall opacity |
There was a problem hiding this comment.
more comments doesnt mean more better, it often means more garbage someone reading has to filter through
| // resolved texture. Cached; the transition shaders are pure crossfades. | ||
| virtual SP<IRendererTexture> renderFitFrame(const SP<IRendererTexture>& src, const CBox& box, eImageFitMode fitMode) = 0; | ||
| // Drop all cached fit-resolved pre-render FBOs (call when a transition starts/ends). | ||
| virtual void clearTransitionCache() = 0; |
There was a problem hiding this comment.
come on man.
// this function eats a banana
void eatBanana();
|
Sorry @vaxerski but I'm gonna have to let this bitrot for a few more weeks up to a few months. I just don't have the time, unfortunately. I'll pick up where i left as soon as i can. As for these setters, that's sloppy on my part. Will review better next time before pushing. Sorry for that. |
|
@markg85 since you said you're short on time, i did the open review bits on top of your current head: https://github.com/UncleJ4ck/hyprtoolkit/tree/rescue-105-transition it collapses |
Companion to: hyprwm/hyprpaper#371
Note: This is written against tag v0.5.4.
Consider this a draft!