Correct code style

still reproduce only for step: Try to use choose image from image selector

still reproduce
eflete commit stack

Alternative flow:
Try to use image manager

It appeared only in case run eflete under gdb.
Today, really I cant reproduce it.

#1 0x0000000000448f53 in _key_press_event_cb (data=0x0, type=56, event=0x33f17a0) at ../../src/bin/ui/shortcuts/shortcuts.c:278
#2 0x00007ffff6a1483f in _ecore_call_handler_cb (func=0x448e30 <_key_press_event_cb>, data=0x0, type=56, event=0x33f17a0) at lib/ecore/ecore_private.h:317
#3 0x00007ffff6a15781 in _ecore_event_call () at lib/ecore/ecore_events.c:518
#4 0x00007ffff6a1e088 in _ecore_main_loop_iterate_internal (once_only=0) at lib/ecore/ecore_main.c:2344
#5 0x00007ffff6a1c2a2 in ecore_main_loop_begin () at lib/ecore/ecore_main.c:1286
#6 0x00007ffff7800002 in elm_run () at lib/elementary/elm_main.c:1248
#7 0x0000000000418ff4 in elm_main (argc=1, argv=0x7fffffffddb8) at ../../src/bin/main.c:334
#8 0x0000000000419083 in main (argc=1, argv=0x7fffffffddb8) at ../../src/bin/main.c:343

Evidently I should accept, because now we can't check working our shaders on nvidea drives.
We will notes it for the future.
Thank you.

thank you @stefan_schmidt for the report

Needn't All examples compiles though make examples.

Please check it again. I built Efl with flags egl without any issues.
Also I saw you do two commits that can influence to generation shader code. It is possible that you got not up to date evas_gl_3d_shaders.x file.
So you can just remove evas_gl_3d_shaders.x file and build efl, it should helps.

	Why don't you just pass "3d" ? Also it is strange to require the name to have 3d, I have been thinking about it, but isn't it possible to actually have an edje function that will return true on an object that contain 3d object and require a scene.

more safety in returned value

Add dummy for 3D engine function that needn't implemented in software_generic

move functionality to efl_canvas3d_scene module

Interface method buffer_map assume returns pointer to pixels, that image contains, but image that used for render scene doesn't contain pixels. We plan get it from FBO (see 1445 - 1470 in file evas_gl_3d.c) and store it in additional image, see line 855, 856 file evas_canvas3d_scene.c.
Can you prompt where I can store need pixels that I will get from glReadPixels?

Remove metions about fxaa

Correct description

Remove useless changes in evas-3d-cube2.c file

Some more renaming.
Remove auxiliary file.

Some more renaming

This example was planned as illustration working render to separate texture.
Now we have FXAA that also use render to additional texture

Honestly, I just realized current ShadeMode

Generally, Shade is somewhat closed to shadowing and coloring.
not color_pick and post_processing.
If this Shade Mode is about Shader, then I really like to rename it Shader Filter or Shader Mode. It's much clear.

Remove wrong copy-paste

Use more suitable name.
Rename e3d_renderer_sampler_targettexture_get to e3d_renderer_sampler_shadowmap_get

Remove useless variables

Update evas-3d-shadow example

Remove API's regard to FXAA from evas.canvas3d.scene.
Use evas.object property anti_alias.

also I attached file with example

Note: stensil -> stencil. I'll add a patch to fix it.

Thank you

Reduce amount of polygons for models

We are implementing FXAA anti-aliasing with help on render to additional texture. So I am not sure that all work fine together. Need more time for test.

Rebase. Changes regard to last Eo syntax

Rebase with master. Correct to last changes of Eo.

There is some issue with render to texture with/without shadow effect. Need make sure that all work correct

restore previous models

clean up models (*.obj files)

In case switching between model and billboard (compile example, run and press key "1") we have wrong displaying of billboard object, see wrong.png

Regard to inline function in evas_3d-unitls.h. In my opinion, it is not need have they inlined, furthermore part of them have long code for inline functions.