• Content count

  • Joined

  • Last visited

  • Days Won


Vousk-prod. last won the day on December 19 2015

Vousk-prod. had the most liked content!

About Vousk-prod.

  • Rank
    Advanced Member
  • Birthday

Contact Methods

  • Website URL

Profile Information

  • Gender
    Not Telling
  • Location
    Strasbourg - France

Recent Profile Visitors

909 profile views
  1. Hi, I have a scene with a renderTargetTexture (a reflectionProbe to be precise). The scene is fully loaded, everything is fine in a calm and peacefull world. Then, I suddenly decide to load new meshes, via SceneLoader.ImportMesh. In the succes callback I'm using scene.executeWhenReady to do some post treatment. And again, everything is fine, the scene goes well, the new meshes live their lifes quietly in a stable universe and the gods are happy. But if at some point I disposed the probe before loading those new meshes, the scene.executeWhenReady is not called! I tried disposing manually the renderTargetTexture inside the probe and the problem also occurs, it is clearly related to renderTagetTexture disposal, as if some flag was forgotten somewhere, leading to scene never ready again. The case where this bug occur is pretty specific so I do not have a repro case in PG for now. But maybe with the information I gave that would make someone here have a "tilt"...
  2. In fact that's a good question. ActionManager is used to manipulate events on meshes, but if you want to catch those advanced events outside a mesh (on scene for instance), I think you have to implement yourself the same mechanims already set in scene.attachControl (which raises the ActionManager events). We should create new "advanced" Observers (longpress, leftpick, etc) in BJS and automatically notify them same time as ActionManager. I'll have a look at this. BTW, I've just noticed that ActionManager's doc is not up to date (eg, it is not explained that the difference between onPickEvent and onPickDownEvent is that onPickEvent is only triggered if mesh is clicked but not dragged, while onPickDownEvent is always trigged when mesh is clicked), I'll fix this.
  3. OK. Issue opened : https://bugzilla.mozilla.org/show_bug.cgi?id=1338704
  4. On Android, except for some rooted devices, the update is automatic, and also on IOS (except for those who have too old device, but Apple do everything they can to "force" people to always buy latest device )
  5. Ok, I will remove it. But it still is strange, why this does not occur if engine is not resized. Do you have any clue?
  6. @Deltakosh I've created the test page for bugzilla, but before posting I'd like you to test something. I noticed that the bug occurs only if engine.resize is called. So I've added a button to de/activate that, but I'm asking if I should remove this button or not. Maybe it could be disturbing for them to debug, or they could say that it is BJS related and not browser related. What do you think ?
  7. Yep... for desktop... but mobile browsers will certainly take some time. And Safari...
  8. Yep, when I saw this line I also find it suspiscious, no alpha can be taken in account with that. I corrected that and it's ok now, PR is on its way. But there still is another last problem with this new screenshot : the final image is upscaled instead of directly rendered at specified size. renderContext.drawImage(engine.getRenderingCanvas(), offsetX, offsetY, newWidth, newHeight); @Deltakosh Is there a way to make the engine render at final size instead of using getRenderingCanvas (where image has current size)
  9. I don't think this is possible for now. Ping @jahow ?
  10. Yup, we can think it is a Chrome bug. But sometimes a strange behavior is not a browser bug but a erroneus way to do something you'll only detect thanks to a browser who do not accept a half working stuff. (I'm absolutly not saying your new screenshot code is wrong, I only open the door for a double check, because it worked before, and I cannot reproduce the bug when I try to manually do the same kind of your new screenshot mecanism)
  11. To submit to bugzilla I need more info. Can I bypass engine resizing in the PG ? Is there a particular process that can be involved in cropped 3d area (maybe camera viewport related) when engine resizing is called ?
  12. It depends... january 2015, when we had a bug in FF with instanced multisubmaterials meshes, it took almost a year for them to consider and debug...
  13. Sorry, in fact, no. The outlines showed by Nabroski was a digression and a side effect of specific alpha usage, but the fact that Chrome does not take alpha in account in new screenshot method whereas it worked in old screenshot and with a standard canvas.dataToURL is a real bug, not solved for now.
  14. Yep, that would make sense, but there is also the fact that if we resize canvas without calling engine.resize(), the bug does not occur (but of course the 3d is then very pixelated). I can't manage to reproduce it on the playground, I presume the PG environment automatically call engine.resize if changes in the layout are detected, I cannot bypass that to test more deeply.
  15. The bug mentionned above seems to happen only on FF. On Chrome the drawing area is not cropped, but above a certain size, there is no 3D at all.