X-Ryl669

Well, I guess it's a usability issue. In your sandbox, try to use a large picture (like 12MPx one). Once clicked, we should have an option to allow more zoom...

The paper: https://arxiv.org/ftp/arxiv/papers/1810/1810.11352.pdf

I don't know how to use Zig on my system (ESP32). I don't know how hard it would be to rewrite the rendering engine in C++ (or at least C).

Reference test code: ```cpp lv_obj_t * img; img = lv_img_create(lv_scr_act()); lv_img_set_src(img, "examples/libs/gif/bulb.gif"); lv_img_set_play_mode(img, LV_IMG_CTRL_PLAY | LV_IMG_CTRL_FORWARD | LV_IMG_CTRL_LOOP); lv_obj_align(img, LV_ALIGN_LEFT_MID, 20, 0); img = lv_img_create(lv_scr_act()); lv_img_set_src(img, "examples/assets/img_star.png"); lv_obj_align(img, LV_ALIGN_LEFT_MID, 100,...

Don't understand the CI failure. Should be ok for review.

What do you mean?

It's based on the previous code, that is a list of decoder context (instance). If the decoder decoded the picture in an internal buffer, it's remembered. The list is filtered...

> * why can't we use the generic `lv_img` for lottie? We can and we are. If one does not need specific "lottie" feature, the usual `lv_img_set_src` works perfectly. However...

> The goal of getting the metadata is to get the size of image, right? And also the frame rate (or delay between frame for variable frame rate format like...

It's not related. Your PR should still apply over this one.