You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

29 KiB

Evas 1.7.99





eina (1.6.0 or better)
freetype (2.1.9 or better)

libX11 + libXext + libXrender
OpenGL2.0 or OpenGL-ES 2.0
libjpeg (6.0 or better)
eet (1.5.0 or better)

XCB SDL esvg libtiff libwebp libgif edb DirectFB evas_generic_loaders

Evas is a clean display canvas API for several target display systems
that can draw anti-aliased text, smooth super and sub-sampled scaled
images, alpha-blend objects much and more.

Evas is designed to be portable to different display systems. Evas uses very
little RAM too (try profiling it in memprof if you want to
know) most of the ram allocated, if you look, is for freetype itself,
image pixel data, and font glyph data. You can't really avoid this, though
evas tries to share this data as much as possible and not duplicate where it
can. Feel free to point me at sensible memory optimizations etc. though :) I
want this baby to be lean, mean tiny, fast and do everything from your
massive multi-cpu desktop with gobs of ram and disk to a tiny watch.

Evas also supports full UTF-8 for text object strings, thus allowing for
full internationalized text strings (if your font gives you all the
characters). I've tested with quite a few fonts and it works quite well.
Though this requires a unicode compatible font with unicode charmap support
(cyberbit is quite good actually as a font). For now Evas draws the fonts
only from left to right, so arabic, hebrew etc. won't display quite right,
direction-wise, but the characters do.


(as root unless you are installing in your users directories):
make install

if you want to know what options to enable
./configure --help

Evas's rendering code assumes a decently optimizing compiler. For
example gcc with -O2 -march=nocona for example (compile for core2 duo
x86 or better). You may choose not to compile for a very modern
architecture, and Evas still has MMX/SSE, NEON and other hand-crafted
assembly, but not for everything, so make use of your compiler
optimizing as much as possible. At least use -O2 or equivalents.

the small dither mask is faster on the ipaq, but is not as good looking. on
desktop machines it makes no speed difference so only use
--enable-small-dither-mask if you really need the speed for low depth
you need at least 1 image loader if you want to load images.
gcc 3.0.x on solaris screws up the jpeg code so erroring out doesn't work.
use gcc 3.2 on solaris.

notes on features (--enable-FEATURE enables it and --disable-FEATURE
disables it, some being enabled or disabled by default or if
dependencies are found):

Also the wayland support (EGL and SHM engines) is considered experimental as
wayland itself is still unstable and liable to change core protocol.
If you use this api, it is possible it will break in future, until this
notice is removed.


this enables the sampling scaler code. this is the fastest image scaling
code, but also the lowest quality. when scaling up pixels will become blocky
and when scaling down you will see shimmering/aliasing artifacts. this is a
speed vs. quality tradeoff


this is the nicest looking scaler that is not that much slower than
tri-linear, but it looks really good.


this uses a 4x4 dither mask instead of 128x128. on desktop boxes these days
(pentium, pentium2, amd etc.) the speed difference is not really measurable,
but the quality of the 128x128 dither mask is quite a lot better. patterns
of dithering are much less noticeable, so it is recommended to not enable
this unless you are struggling for speed. the compaq ipaq for example shows
a slowdown with this large a dither mask so enabling a small dither mask is
recommended unless you really want to forgo the speed.


this is a faster alternative to the small or large dither masks above.
this dithers only on an alternating-line basis. this only provides 1
intermediate "dither" level whose odd and even pixels alternate
between the 2 closest colors available, but it is very fast. almost as
fast as no dithering. quality though will not be as good as small or
default "large" dither masks.


this disables dithering entirely. this is the fastest option, but the
lowest quality. not suggested in general unless you are really in need
of an extra few percent speed and are willing to have fairly awful
quality. but in general this is the standard rendering for most
"realtime graphics" if it has to drop to lower bit-depths, so it's
not anything unusual. just in the evas world the quality is considered
poor enough to be discouraged as evas's internal rendering is so much
higher quality.


All engines can be compiled statically inside in order to
reduce runtime time at dynamically loaded modules. All you have to do
is to enable your chosen modules with "=static" suffix. If they depend
on software_generic (most do), you need that as well. Examples:

./configure --enable-static-software-generic --enable-software-xlib=static


this enables the software x11 rendering engine that renders to X drawable
targets using highly optimized software routines. there is no hardware
assist here. this engine requires X11 to be installed to build (and run).
This is a good generic engine that is fast and can run in X for good
development and debugging purposes.


this enable the software xcb rendering engine. It allows the same
features than the software xlib engine. It require the XCB and XCBImage
libraries. For the test programs, XCBICCCM is also needed. It is not
recommended to use this as it's experimental and will create problems
with both ecore_evas and enlightenment itself.


this is the software framebuffer driving engine. this uses the linux
framebuffer device (/dev/fb{X}) and will currently just inherit the current
framebuffer settings on the fb device and use them to run in. this engine is
almost fully functional except for the fb management itself. this engine is
specifically geared towards people writing minimalist display systems for
embedded devices such as the ipaq, zaurus, etc. it also scales up to high-res
desktop systems as


this is the direct fb engine that uses direcftb ( on
linux to access the framebuffer with (or maybe without) acceleration. for
people making set-top boxes or just wanting an alternative to X this is
really good. it may also be useful for embedded devices supported by
directfb that offer acceleration (otherwise the fb driver will likely be
faster). as such this engine is in relative disrepair and is not
maintained. use with great care.


this enables the memory buffer rendering engine. this engine renders
to a region of memory that is considered to be a 32bit ARGB buffer of
pixels, allowing the results of rendering to be directly read out or
used again for other purposes.


this is the opengl engine. it is intended for an x11 target (via xlib)
rather than framebuffer (even if you use EGL, the EGL flavor is
expected to be an x11 one). it is a full opengl based rendering engine
with all rendering implemented as a texture + triangle pipeline and
more. it also supports opengl-es2.0 and is reliant on modern opengl2.0+
shader support. this engine also supports the native surface api for
adopting pixmaps directly to textures for compositing.

some environment variables that control the opengl engine are as

export EVAS_GL_INFO=1
set this environment variable to enable output of opengl information
such as vendor, version, extensions, maximum texture size etc. unset
the environment variable to make the output quiet again.

set this environment variable to enable dumping of debug output
whenever textures are allocated or freed, giving the number of
textures of each time and how many kb worth of pixel data are
allocated for the textures. unset it again to stop this dumping of

set this environment variable to enable the gl engine to try and
ddelete the window surface, if it can, when told to "dump resources"
to save memory, and re-allocate it when needed (when rendering
occurs). unset it to not have this behavior.

set this environment variable to the maximum number of rectangles
applied to a rendering of a primitive that "cut away" parts of that
primitive to render to avoid overdraw. default is 512. unset it to use
defaults, otherwise set N to the max value desired or to -1 for
"unlimited rectangles".

set the maximum number of parallel pending pipelines to N. the
default number is 32 (except on tegra2 where is it 1). evas keeps 1 (or more)
pipelines of gl draw commands in parallel at any time, to allow for merging
of non-overlapping draw commands to avoid texture binding and context
changes which allows for more streamlining of the draw arrays that are
filled and passed to gl per frame. the more pipelines exist, the more
chance evas has of merging draw commands that have the same modes,
texture source etc., but the more overhead there is in finding a
pipeline slot for the draw command to merge into, so there is a
compromise here between spare cpu resources and gpu pipelining. unset
this environment variable to let evas use it's default value.

set the size (width in pixels) of the evas texture atlas strips that
are allocated. the default is 1024. unset this to let evas use its
default. if this value is larger than the maximum texture size, then it
is limited to that maximum size internally anyway. evas tries to
store images together in "atlases". these are large single textures
that contain multiple images within the same texture. to do this evas
allocates a "wide strip" of pixels (that is a certain height) and then
tries to fit all images loaded that need textures into an existing
atlas texture before allocating a new one. evas tries a best fit
policy to avoid too much wasting of texture memory. texture atlas
textures are always allocated to be EVAS_GL_ATLAS_ALLOC_SIZE width,
and a multiple of EVAS_GL_ATLAS_SLOT_SIZE pixels high (if possible -
power of 2 limits are enforced if required).

this is exactly the same as EVAS_GL_ATLAS_ALLOC_SIZE, but for
"alpha" textures (texture used for font glyph data). it works exactly
the same way as for images, but per font glyph being put in an atlas
slot. the default value for this is 4096.

set this to limit the maximum image size (width) that will be
allowed to go into a texture atlas. if an image exceeds this size, it
gets allocated its own separate individual texture (this is to help
minimize fragmentation). the default value for this is 512. if you set
this environment variable it will be overridden by the value it is set
to. the maximum value possible here is 512. you may set it to a
smaller value.

this is the same as EVAS_GL_ATLAS_MAX_W, but sets the maximum height
of an image that is allowed into an atlas texture.

this sets the height granularity for atlas strips. the default (and
minimum) value is 16. this means texture atlas strips are always a
multiple of 16 pixels high (16, 32, 48, 64, etc...). this allows you
to change the granularity to another value to avoid having more
textures allocated or try and consolidate allocations into fewer atlas
strips etc.

if this environment variable is set, it disabled support for the SEC
map image extension (a zero copy direct-texture access extension that
removes texture upload overhead). if you have problems with dynamic
evas images, and this is detected by evas (see EVAS_GL_INFO above to
find out if its detected), then setting this will allow it to be
forcibly disabled. unset it to allow auto-detection to keep working.


this enables the opengl-es 2.0 flavor of opengl (as opposed to desktop
opengl) when building evas's gl-x11 engine above. this will be needed
if you are building evas for opengl-es 2.0 enabled embedded devices.
evas works on several opengl-es 2.0 compliant gpu's and gains more
testing and optimization regularly. it is also capable of
texture-from-pixmap support in opengl-es like it is in desktop opengl.


this tells evas that you are building the gl-es engine for a
shader-compiler "sgx style" opengl-es 2.0 implementation. this is
where the shader compiler is provided at runtime and can accept the
shader glsl source and work


this tells evas that you have an s3c6410 style opengl-es
implementation that has an offline shader compiler and that needs
pre-compiled shader binaries (provided with evas). this has not been
tested in quite a while as the drivers and environment for this system
have gone missing


windows gdi based engine for evas


windows direct-draw engine for evas


evas direct3d engine (experimental)


this is the sdl engine that uses sdl library ( This
library should work on many operating system. the buffer is
software-rendered with evas's default software rendering core.


opengl (and opengl-es2.0) rendering engine that uses sdl as the front
end interface. see --enable-gl-x11 etc. for information.


this enabled the c code. you can actually build the code without the c
fallback code and only have the mmx routines for example. it is suggested to
always use this regardless unless you have some definite size issues with the


this enables the mmx optimized routines. this works for pentium, pentium2,
pentium3, pentium4, athlon and duron processors. it can get quite
considerable speedups, souse it if you can. ppc owners just have to live with
the c fallback functions unfortunately as no one has provided any ALTIVEC asm
routines yet. :) arm owners will also have to rely on the c fallback
routines as i haven't managed to come up with any arm assembly that actually
can beat the c code (when compiled with all optimizations) in speed.


this enables sse optimizations available in he pentium3 and 4 cpus (not
athlon and duron or pentium 2 or pentium cpu's). ppc owners just have to
live with the c fallback functions unfortunately as no one has provided any
ALTIVEC asm routines yet. :) arm owners will also have to rely on the c
fallback routines as i haven't managed to come up with any arm assembly that
actually can beat the c code (when compiled with all optimizations) in speed.


this enables sse3 optimizations available in the Intel Pentium4, Core, Xeon,
and Atom processors, as well as the AMD Athlon64, Phenom, Opteron, and Turion


This enables support for the Arm Cortex-A8 and later Neon register
set. In particular it will use neon optimized code for rotations and
drawing with the software engines. Open GL based renderers will gain
nothing from the use of neon.

To use neon with gcc-4.4 you need a post-2009 gcc and options
something like: -mcpu=cortex-a8 -mfloat-abi=softfp -mfpu=neon
Note that this slightly slows down non-optimized parts of evas but
the gains in drawing are more then worth it overall.

This is enabled by default, and turns off if a small test program is
unable to compile.

Performance is at least 50%, and in some real world tests approaches

If you have any issues with neon, please report them to either the
edevel mailing list or Brett Nash <>


this enables the loader code that loads png files using libpng. there may be
call for embedded devices later that have custom written small image
loaders that uses less disk space than libpng to load custom format images.
for now this is the only loader so you may as well include it.


this enables the loader code that loads jpeg files using libjpeg. this
loader also supports load options to pre-scale jpeg images down to
provide much faster load times while also getting downscaling by 1/2,
1/4 or 1/8th the size in each dimension for "free". with an added
patch to libjpeg7, it can also fast-decode a specific region of a jpeg
file (without the patch it take a slow-path to do this).


edb image loader- can load images inside edb database files. not very
useful as edb itself is no longer used by enlightenment. may be
removed at some point, so unless you have a burning need for this,
don't use edb files to store image data and rely on this loader


loads image data from eet files. eet files are the backing for edje
storage, so this is needed for edje to work. it is very useful as it
can load an image from anywhere in the eet archive by key value so eet
files are like "zip" files where you can pack a whole lot of image and
other data together and just pick out the pieces you need at runtime.
requires the eet library.


gif image loader. gif is an obsolete format, but due to its longevity,
sitll has lots of existing data around.

NOTE: evas has no notion of time, thus animated gif file are not


ppm/pnm/pgm image loader that can load the "pnm" style image format.
not very common, but the files are simple raw RGB, greyscale image or
bitmap data in binary or ascii format


this loader can load svg files via esvg (thus it is a dependency). This
loader supports load options to set the dpi to decode the svg at etc. which
can then be used to create scalable images that scale to any size without
becoming blocky or blurry, if the source is an svg file.

Esvg can be found here:

Install (in that order):



this loader uses libtiff to load tiff image files


this loader uses libwebp to load webp image files


this is an xpm format image loader. xpm format images are ascii files
that look like c/c++ source code that contain images. these files are
old-fashioned unix+x11 images you may encounter, but are inefficient
for storage and decoding and have been superseded by png files in
almost every way


this enables the bmp image format loader. note that there seems to be
a disagreement on 32bit bmp format images where alpha channels are
concerned and you may run into issues with bmps generated by the gimp
that have alpha channels. there is a problem where they don't seem to
be spec-conformant.


this loader load tga format files. these files are very old-fashioned
but found often in the 3d graphics world.


this loader will execute a given binary to decode an image and read
the resulting image data via a shared memory segment, a mmaped file or
stdout. it uses the command-line to pass the filename and any load
parameters, and reads stdout from the loader binary to get metadata like
width, height, alpha channel flag and location of pixel data. this
loader has no dependencies as the binaries run are to be found in
PREIFX/lib/evas/utils and are named evas_image_loader.EXTENSION where
.EXTENSION is replaced by the filename extension to be decoded. if
this binary does not exist then evas_image_loader (with no extension) is
tried as a last fallback allowing it to handle "all cases".

since this loader doesn't use any libraires, it relies on runtime
dependencies and executables existing in the utils directory. note that
images loaded via this mechanism will have slower load times due to the
overhead of execution of another binary, but any instability in the
loaders themselves will not affect the application using evas.

this also means that licenses such as GPL for the binaries in this
utils directory do not affect evas and the applications or libraries
using evas.

there is a separately released evas_generic_loaders package which
builds stand-alone binaries that can do this style of decoding for for
evas. this package currently handles XCF, PDF, PS, RAW, SVG (via
librsvg) and video formats (via gstreamer).


this loader can load font (ttf) files directly from eet archives like
the eet image loader. requires the eet library


this enables an optimized yuv (yv12 601 colorspace) to ARGB32
converter in evas


the most common converter you'll want for 16bpp. this means 5 bits for red,
6 bits for green and 5 bits for blue are used.


this is a converter for what many people know as "15 bit" color. you might
want to enable this for X output as it used to be common to find many cards
that do this.


this converter outputs to 12bit packed (int 16 bit WORDS).


this converter was written specifically for the ipaq (and may apply to
similarly configured devices) because it lies about its screen depth. it
says it is 16bit 565 (that means 5 upper bits of the WORD are red, the next 6
bits are for green abd the next 5 for blue) but in fact only the upper 4
bits of each color component (red green and blue) are significant and work,
so effectively the display is 12 bits of color, not 16, but padded out to
fill 16bits, with unused bits in the color masks. X on the ipaq advertises
it as a full 16bpp 565 display (i can't remember what the linux framebuffer
advertised it as) and so many lumps of code can be fooled into rendering
data badly because they think the output will look as the expect. This
renderer assumes the upper 4 bits fo each color primitive only are
significant and renders accordingly. this produces nice quality images on
the ipaq and even still works in 16bpp 565 on your pc. it is highly
recommended to use this renderer if your target is an ipaq or your device
displays similar qualities of the ipaq for display purposes.


this enables the 16bpp converters to run with 0 degrees rotation - this is
normal display and you should really include this (though it is optional if you
only ever want to do portrait mode - perhaps like on an ipaq embedded device)


this enables the portrait mode (270 degree rotation) converters for 16bpp.
this is the standard display mode for things like pocketpc on the ipaq and
the zaurus etc. this is an optimized part of the rendering pipeline to allow
portrait display with a much lower overhead than doing it through X.


same as --enable-convert-16-rgb-rot-270 but for 180 degrees


same as --enable-convert-16-rgb-rot-270 but for 90 degrees


this converts evas's 32bit ARGB to 24bit RGB packed format for output
if needed


this converts evas's 32bit ARGB to 24bit packed BGR format for output
if needed


32bit RGB output conversion support. byteswapping compared to evas's
native colorspace


conversion (reduces to a memory copy) from evas's native colorspace to
the same color format.


copies without rotation evas's native image format


copies evas's native ARGB32 pixels but at a rotation of 270 degrees.


same as --enable-convert-32-rgb-rot-270 but for 180 degrees


same as --enable-convert-32-rgb-rot-270 but for 90 degrees


a special colorspace handler for 18bit color packed into 24bit output
(where only 6 bits per r, g and b byte are used). the only known
platform that did this was the motorola esx based phones that used
qtopia originally and have open ezx ports for them.


enable 8bit gray to 1 bit black & white converter


8bit grey to 16 level grayscale converter


8bit grey to 64 level grayscale converter


enable converter from 32bit ARGB to 8bit color "332" colorspace (3bits
red, 3 bits green, 2 bits blue)


enable converter from 32bit ARGB to 8bit color "216" "websafe"
colorspace (6 values for red, 6 for green and 6 for blue - 6x6x6 being
216 colors).


same as convert-8-rgb-332 but 2 bits red, 3 green, 2 blue


same as convert-8-rgb-332 but 2 bits red, 2 green, 2 blue


same as convert-8-rgb-332 but 2 bits red, 2 green, 1 blue


same as convert-8-rgb-332 but 1 bit red, 2 green, 1 blue


same as convert-8-rgb-332 but 1 bit red, 1 green, 1 blue. this is the
lowest sized colorspace supported for rgb (3bits, 8 color).


this enables pthread support in evas so multiple threads may run
internally for parallel rendering, loading etc.


this provides the ability for evas to have an asynchronous event
notification pipe to provide events when background threads are done
with tasks, like pre-loading image files


evas can load images (preload) in the background using a thread
if you ask it to, and provide events when done. this goes hand-in-hand
with --enable-pthreads and --enable-async-events. you really want all
of these available.


this enables a multiple-thread renderer that divides the rendering
into N regions (1 per core) to speed up rendering in software when you
have multiple cpu cores.


this enables fontconfig support for loading font files by using
generic fontconfig font names and styles. you really should use this
by default on any linux/unix platform for universal font support.


this enables support for the fribidi library to have right to left and
left to right font rendering so languges such as arabic, hebrew and
other "RTL" langauges display properly.


this enables support for the harfbuzz shaping library for complex
shaping support in arabic, hindi and other similar languages.


this enables support of complex line breaking rules using liblinebreak.


this allows you to enable and disable evas's extra magic number
checks. these allow better stability with runtime object magic
"number" checks to make sure you are accessing a real object in memory
of the right type, and will avoid doing "bad things" if they detect
the wrong object type being passed in. if you are absolutely sure your
system has no bugs in accessing objects of the wrong type with the
wrong calls, you can gain some small performance by disabling this.


For the arm optimizations you want to try:
export CFLAGS="-O2 -march=armv5te -mcpu=arm1136jf-s -fomit-frame-pointer"

To enable the async renderer compile with:
and also runtime set this environment variable:
export EVAS_RENDER_MODE=non-blocking

For compilation with MinGW, fnmatch.h is probably missing. That file can be
found here:
It should be installed in the mingw include directory.