quirks, requests, questions and answers

[taumelSpectator]

@Mark
Where do you prefer getting bugs and glitches reported? Using the issue tracker, here, doesn’t matter?

Anyway, i just stumbled across:

ted2:
a) No umlauts

Fullscreen OS X 10.11.5:
a) Once you return from fullscreen mode, the definitions (like Mission Control) for active edges on the right side aren’t working anymore (which generally works).

[wiebowParticipant]

Mark wants to use the issue tracker.

[coconParticipant]

Mark accepts contibutions on github repository?

[taumelSpectator]

issue tracker

[taumelSpectator]

A few things from the last time i played around with m2 (OS X, 1.0.1/1.0.2[ogg]):

Btw. where do you post feature requests (thread, issue tracker)?

Requests:
a) Support for loading a folder structure directly, without the #Import, like in Max. Would be more flexible.

b) DrawImage/Load would nice if you also could provide a parameter for atlas/animations/packed textures (like, width, height, [amount]) to access the sprites within the atlas. Currently you need to split all the data into several images in m2 (isn’t this relevant performance related?, didn’t have time to check out the source [yet]).

c) FLAC support.

Question:
a) How do you check the Mouse-Events/Buttons States in all different ways? I got it working one way (event with test on int) but not the others, only checked out the doc, not source/sdl).

b) Does DrawImage support Clamp as a Warp Mode?

Quirks:
a) Wasn’t able to ‘just’ replay oggs with the proper sample rate (also in audiotest.monkey”.

b) In some cases you needed to kill the task in order to exit a debug release.

c) Mostly m2 runs fine but in rare cases there can be a hickup and it stutters for a short moment.

d) Some fullscreen modes don’t work as expected (wrong mode, stretched only over a part of the screen).

e) PixMap operations are quite slow.

[Simon ArmstrongParticipant]

Taumel, if you can post some code that reproduces some of these issue you would be making a valuable contribution.

I think pixmap performance is fine and images created as children of atlas are also fine.

There is i think only clamping, maybe wrapping but no transparent border color for scaled sprites you need a gutter of 0x00000000 around all cells of 2^n where n is your max mip map generation.

Or perhaps tweak the mojo shaders where EGL spec expects a lot of this work to now be done.

If you are not scaling sprites then don’t use the mipmap flag but you will still need a single gutter of alpha 0 if you want to position in subpixel space.

I am also unsure about pre multiplied stuff and have been meaning to play with the shaders to fill in my knowledge of this stuff.

I personally would like to operate with textures at gl level and wondering if I can access them not from image handle but through the GL uniform setting where one could call glDrawPixels with abandon.

[taumelSpectator]

@mouse
Any idea?

@clamp
The filtering mode doesn’t matter, line/interpolation without setting up the texture accordingly. Tile and offset support for the images would be nice to have together with wrap modes clamp and repeat.

@ogg
‘Audiotest’ It was a matter of data, the sounds were replayed too low.

@hickups
They show up in a number of content, if you watch long enough.

@debug
Happened f.e. when building for debug accidently and more images were generated/drawn.

@fullscreen
Happend with smaller resoutions than 1280×800 here (=2440×1800). ‘Super.New(title,width,height,WindowFlags.Fullscreen)’ hey hey
Sometimes you want this behavior (but centered in the middle of the screen then). For the rest it’s not the proper screen size. Ideally you might want go windowed with this size, if you change to fullscreen use that size (same size without further defs) or go to that res but show the content in its original size (flag, [wanted size]). I’m aware of projection. It’s very nice having low level access but the charm more often is about how easy, fast and fun it is to use something.

@pixmaps
I think r/w access could be faster, also depends on the hardware you’re looking at.

[taumelSpectator]

Hmm, caused bus errors, interesting, could fix it.

[taumelSpectator]

@bugs
When trying to use larger fonts (f.e. height=144), m2 reports a memory access violation. I’m on 1.0.1 but it happens in 1.0.2 too.

[taumelSpectator]

@requests gfx
Hmm, i think some blendmodes are missing (like substraction, difference, hard ligths). Generally Bloom and AA are missing. As soon as you deals with lights, hdr and tone mapping would be cool too.

[nobuyukiParticipant]

Texture size is limited by your graphics card.  There is an attribute available in openGL called MAX_TEXTURE_SIZE which you can refer to using the glGet() function from gles20.

See:  https://www.khronos.org/opengles/sdk/docs/man/xhtml/glGet.xml

Misread the original post.  Thought the size was extremely large….

[taumelSpectator]

Yeah, it’s not like a 144m high. :O) Otherwise m2 deals with 8k textures nicely.

[taumelSpectator]

@blog
It’s a good idea to concentrate on less right now. Apart from the module support, there is desktop, the docs, ted2, enhancing gfx and sfx, … otherwise it’s likely turning into a mess with nothing done right. Most people dislike dealing with unfinished stuff, no matter which platform they prefer.

I hope that the tweaks for ted2 also include many fixes.

@questions
Btw. is there an option to start the posts with text instead of visual by default?
Is the monkey gfx on the upper left the official m2 gfx? Is it available in a vector format (and an url for the old one)?

I’m currently using this one as the standard app icon was kind of boring.

[taumelSpectator]

+) I like the ‘hot pink’ in the icon.
-) I dislike that ted2 is still low res, major bummer.
:

My stuff compiles fine, meanwhile i also learned to enjoy compiling monkey2.

@todo
sfx: Take a look at the licenses for fmod and wwise.

[taumelSpectator]

@blog

retina:
Retina displays are around since more than 4/6 years. I really would like to see proper out of the box support. The low res strains the eyes and makes ted2 look dated. Your first contact with m2 is via ted2, so, it matters.

fullscreen:
#1 Choosing resolutions smaller than 1280×800 (here = 2880×1800) results to a) wrong clipping (visible plopping) and b) content which isn’t spread over the whole screen. Instead it seems to be the chosen resolution from the upper/left only partly covering a larger screen. #2 Returning from fullscreen mode to the OS results in broken active edges definitions on the right.

[Author]

Posts