@hermet@JSUYA we had this conversation a few times already... I sent prs, that depend on each other (#816 depends on #812). We have a few prs regarding gradients and I tried to split them to be as small as possible. I know that this doesn't have the highest priority, so you may review them in some time, so I wanted to send all of them now to be able to concentrate on something else. Returning to all of these calculations after a break and deciding what change should be sent as the next one is very time consuming.
@mgrudzinska neat and clean. #812 has been applied. please update the radial case as well. thanks.
@hermet for now in capi we have 'tvg_paint_transform' + 'tvg_paint_get_transform' and 'tvg_gradient_get_transform' + ''tvg_gradient_set_transform'' Can we change 'tvg_paint_transform' -> 'tvg_paint_set_transform' ?
@mgrudzinska Yes, it seems has to be. Go Go change please.
@ifire:mozilla.org@projectitis please have a look at this Samsung/thorvg#594, static loader for png/jpg modules has been submitted. thanks.
@hermet as @mihashco noticed, the explicit conversion is needed in the case when the bitwise or is used on engine types. since we release capi today, I sent pr in which I've changed the args type. please take a look and decide how we want to deal with it.
I just came across the repo! I was in search for vg renderer for months now (was using nanovg but it was slow for my purposes). Thank you very much. I will try to make a backend for it on bgfx. But I have a question, is performance suitable for games for maybe 100-200 basic dynamic objects on screen, dynamic in terms of movement and shape changes so tesselation caching will not be very efficient?
how does the gl backend work? Does it use stencil buffer for clipping (like what rive-sokol does for example)?
I don't think the GL renderer actually supports this yet from what I see
@kariem2k hi, right. it actually doens't support gl, it's under the development. somewhat sw raster could be better than gl (it's up to how the screen is composed.)
@kariem2k in case of gl , if you have your own native rendering system, it would lose the performance when switching gl contexts between your main renderer and vector graphics renderer (thorvg).
@kariem2k in terms of that scenario, sw raster is more efficient. if you concern the performance, in the end you need to prototype & verify it.
@kariem2k i assume, the number of objects (100 ~ 200) won't be issued much, instead the higher screen resolution is more criticial to the performance.
@kariem2k i don't know the rive-sokol at this moment, you can check https://www.youtube.com/watch?v=U0y0afuNW-8 video I recorded rive animation with thorvg backend on the standard laptop (old one, not high spec machine).
@kariem2k feel free to ask me further questions :)