<div dir="ltr">Hi Christian,<div><br></div><div>I tried using the approach, the artifacts are gone. But for some videos</div><div>the output quality has reduced. The quality for such videos is somewhere</div><div>between nearest neighbor and linear interpolation.</div><div><br></div><div>Regards,</div><div>Nayan</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Jul 21, 2016 at 7:48 PM, Christian König <span dir="ltr"><<a href="mailto:deathsimple@vodafone.de" target="_blank">deathsimple@vodafone.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div text="#000000" bgcolor="#FFFFFF"><span class="">
<div>Am 21.07.2016 um 16:05 schrieb Nayan
Deshmukh:<br>
</div>
</span><blockquote type="cite">
<div dir="ltr">Hi Christian,
<div><br>
</div><span class="">
<div>Yes, that is for pixel center adjustment. </div>
<div><br>
</div>
<div>let me give you an example, for lanczos I need frac(x)
where x is the original </div>
<div>coordinate before scaling. To calculate that first I
subtract half_pixel and then</div>
<div>multiply by the original surface size, which gives me the
original coordinate.</div>
<div><br>
</div>
<div>eg. if the coordinate before scaling was 24.5 (total size
300) after 2x it becomes</div>
<div>49. When the frag shader is executed we get 49.5/600 as the
coordinate so what </div>
<div>I do is 49.5/600 - 0.5/600 = 49/600 and then multiply it
with 300 to get 24.5 the </div>
<div>original coordinate.</div>
</span></div>
</blockquote>
<br>
Well in your case the coordinates are always between 0.0 and 1.0, so
scaling doesn't affect the coordinate.<br>
<br>
You could take a look at how I did that in the weave shader:<br>
1. In the vertex shader I use 0..width instead of 0..1 for the range
(from create_vert_shader() in vl_compositor.c):<br>
<br>
* o_vtop.x = vtex.x <br>
* o_vtop.y = vtex.y * tmp.x + 0.25f <br>
<br>
2. Then in the fragment shader I just need to do the following to
get the original coordinate to sample from:<br>
* t_tc.y = (round(i_tc.y - 0.5) + 0.5) / height * 2<br>
<br>
I use 0.25 and "height * 2" here because the top/bottom fields are
always halve the height and shifted a bit up/down.<br>
<br>
For your case that should just be:<br>
<br>
o_vtex.x = i_vpos.x * video_width<br>
o_vtex.y = i_vpos.y * video_height<br>
<br>
In the vertex shader and then:<br>
<br>
t_tc.x = (round(i_tc.x - 0.5) + 0.5) / video_width<br>
t_tc.y = (round(i_tc.x - 0.5) + 0.5) / video_height<br>
<br>
In the fragment shader to get the correct coordinate. No need to
actually mess with the destination sizes here.<br>
<br>
Regards,<br>
Christian.<div><div class="h5"><br>
<br>
<blockquote type="cite">
<div dir="ltr">
<div><br>
</div>
<div>Regards,<br>
</div>
<div>Nayan.</div>
<div>
<div class="gmail_extra">
<div class="gmail_quote">On Thu, Jul 21, 2016 at 7:20 PM,
Christian König <span dir="ltr"><<a href="mailto:deathsimple@vodafone.de" target="_blank">deathsimple@vodafone.de</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<div bgcolor="#FFFFFF">
<div><span>
<blockquote type="cite">
<div><br>
</div>
<div>This seems to be the reason for the
artifacts.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<div>
<div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
+ ureg_SUB(shader,
ureg_writemask(t_array[0],
TGSI_WRITEMASK_XY),<br>
+ i_vtex, half_pixel);<br>
</blockquote>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>On debugging I found that after removing
this ^^^ instruction the artifacts are gone.</div>
<div>Not sure why is this happening but the
filter is working fine.</div>
<div><br>
</div>
<div>Any ideas Christian?</div>
</blockquote>
<br>
</span> Could it be that your values run out of the
representable numeric range? Otherwise I run out of
ideas as well.<br>
<br>
Additional to that I'm not 100% sure I get what are
you trying to do here. Is that for the pixel center
adjustment?<br>
<br>
Regards,<br>
Christian.
<div>
<div><br>
<br>
Am 20.07.2016 um 14:02 schrieb Nayan Deshmukh:<br>
</div>
</div>
</div>
<div>
<div>
<blockquote type="cite">
<div dir="ltr">Hi Christian,
<div><br>
</div>
<div>Thanks for the review.</div>
<div><br>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Tue, Jul 19,
2016 at 4:58 PM, Christian König <span dir="ltr"><<a href="mailto:deathsimple@vodafone.de" target="_blank">deathsimple@vodafone.de</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><span>Am
18.07.2016 um 21:55 schrieb Nayan
Deshmukh:<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
v2: avoCould it be that your values
run out of the representable numeric
range?iding dividing by zero when
calculating lanczos<br>
<br>
Signed-off-by: Nayan Deshmukh <<a href="mailto:nayan26deshmukh@gmail.com" target="_blank">nayan26deshmukh@gmail.com</a>><br>
</blockquote>
<br>
</span> That looks much better, but
there are still quite a bunch of
artifacts.<br>
<br>
Take a look at the attached screenshots.
good.jpg was created with hqscalling=0,
bad with hqscalling=7.<br>
<br>
Especially on the left side we have
lines from top to bottom where there
shouldn't be any.<br>
<br>
Regards,<br>
Christian.
<div>
<div><br>
<br>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
---<br>
src/gallium/auxiliary/Makefile.sources
| 2 +<br>
src/gallium/auxiliary/vl/vl_lanczos_filter.c
| 447 +++++++++++++++++++++++++++<br>
src/gallium/auxiliary/vl/vl_lanczos_filter.h
| 63 ++++<br>
3 files changed, 512
insertions(+)<br>
create mode 100644
src/gallium/auxiliary/vl/vl_lanczos_filter.c<br>
create mode 100644
src/gallium/auxiliary/vl/vl_lanczos_filter.h<br>
<br>
diff --git
a/src/gallium/auxiliary/Makefile.sources
b/src/gallium/auxiliary/Makefile.sources<br>
index e0311bf..4eb0f65 100644<br>
---
a/src/gallium/auxiliary/Makefile.sources<br>
+++
b/src/gallium/auxiliary/Makefile.sources<br>
@@ -330,6 +330,8 @@ VL_SOURCES :=
\<br>
vl/vl_deint_filter.h \<br>
vl/vl_idct.c \<br>
vl/vl_idct.h \<br>
+ vl/vl_lanczos_filter.c \<br>
+ vl/vl_lanczos_filter.h \<br>
vl/vl_matrix_filter.c \<br>
vl/vl_matrix_filter.h \<br>
vl/vl_mc.c \<br>
diff --git
a/src/gallium/auxiliary/vl/vl_lanczos_filter.c
b/src/gallium/auxiliary/vl/vl_lanczos_filter.c<br>
new file mode 100644<br>
index 0000000..7c69555<br>
--- /dev/null<br>
+++
b/src/gallium/auxiliary/vl/vl_lanczos_filter.c<br>
@@ -0,0 +1,447 @@<br>
+/**************************************************************************<br>
+ *<br>
+ * Copyright 2016 Nayan Deshmukh.<br>
+ * All Rights Reserved.<br>
+ *<br>
+ * Permission is hereby granted,
free of charge, to any person
obtaining a<br>
+ * copy of this software and
associated documentation files
(the<br>
+ * "Software"), to deal in the
Software without restriction,
including<br>
+ * without limitation the rights
to use, copy, modify, merge,
publish,<br>
+ * distribute, sub license,
and/or sell copies of the
Software, and to<br>
+ * permit persons to whom the
Software is furnished to do so,
subject to<br>
+ * the following conditions:<br>
+ *<br>
+ * The above copyright notice and
this permission notice (including
the<br>
+ * next paragraph) shall be
included in all copies or
substantial portions<br>
+ * of the Software.<br>
+ *<br>
+ * THE SOFTWARE IS PROVIDED "AS
IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS<br>
+ * OR IMPLIED, INCLUDING BUT NOT
LIMITED TO THE WARRANTIES OF<br>
+ * MERCHANTABILITY, FITNESS FOR A
PARTICULAR PURPOSE AND
NON-INFRINGEMENT.<br>
+ * IN NO EVENT SHALL VMWARE
AND/OR ITS SUPPLIERS BE LIABLE FOR<br>
+ * ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF
CONTRACT,<br>
+ * TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH
THE<br>
+ * SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.<br>
+ *<br>
+
**************************************************************************/<br>
+<br>
+#include <stdio.h><br>
+<br>
+#include "pipe/p_context.h"<br>
+<br>
+#include "tgsi/tgsi_ureg.h"<br>
+<br>
+#include "util/u_draw.h"<br>
+#include "util/u_memory.h"<br>
+#include "util/u_math.h"<br>
+#include "util/u_rect.h"<br>
+<br>
+#include "vl_types.h"<br>
+#include "vl_vertex_buffers.h"<br>
+#include "vl_lanczos_filter.h"<br>
+<br>
+enum VS_OUTPUT<br>
+{<br>
+ VS_O_VPOS = 0,<br>
+ VS_O_VTEX = 0<br>
+};<br>
+<br>
+static void *<br>
+create_vert_shader(struct
vl_lanczos_filter *filter)<br>
+{<br>
+ struct ureg_program *shader;<br>
+ struct ureg_src i_vpos;<br>
+ struct ureg_dst o_vpos,
o_vtex;<br>
+<br>
+ shader =
ureg_create(PIPE_SHADER_VERTEX);<br>
+ if (!shader)<br>
+ return NULL;<br>
+<br>
+ i_vpos =
ureg_DECL_vs_input(shader, 0);<br>
+ o_vpos =
ureg_DECL_output(shader,
TGSI_SEMANTIC_POSITION,
VS_O_VPOS);<br>
+ o_vtex =
ureg_DECL_output(shader,
TGSI_SEMANTIC_GENERIC, VS_O_VTEX);<br>
+<br>
+ ureg_MOV(shader, o_vpos,
i_vpos);<br>
+ ureg_MOV(shader, o_vtex,
i_vpos);<br>
+<br>
+ ureg_END(shader);<br>
+<br>
+ return
ureg_create_shader_and_destroy(shader,
filter->pipe);<br>
+}<br>
+<br>
+static void<br>
+create_frag_shader_lanczos(struct
ureg_program *shader, struct
ureg_src a,<br>
+ struct
ureg_src x, struct ureg_dst
o_fragment)<br>
+{<br>
+ struct ureg_dst temp[8];<br>
+ unsigned i;<br>
+<br>
+ for(i = 0; i < 8; ++i)<br>
+ temp[i] =
ureg_DECL_temporary(shader);<br>
+<br>
+ /*<br>
+ * temp[0] = (x == 0) ? 1.0f :
x<br>
+ * temp[7] = (sin(pi * x) *
sin ((pi * x)/a)) / x^2<br>
+ * o_fragment = (x == 0) ?
1.0f : temp[7]<br>
+ */<br>
+ ureg_MOV(shader, temp[0], x);<br>
+ ureg_SEQ(shader, temp[1], x,
ureg_imm1f(shader, 0.0f));<br>
+<br>
+ ureg_LRP(shader, temp[0],
ureg_src(temp[1]),<br>
+ ureg_imm1f(shader,
1.0f), ureg_src(temp[0]));<br>
+<br>
+ ureg_MUL(shader, temp[2], x,<br>
+ ureg_imm1f(shader,
3.141592));<br>
+ ureg_DIV(shader, temp[3],
ureg_src(temp[2]), a);<br>
+<br>
+ ureg_SIN(shader, temp[4],
ureg_src(temp[2]));<br>
+ ureg_SIN(shader, temp[5],
ureg_src(temp[3]));<br>
+<br>
+ ureg_MUL(shader, temp[6],
ureg_src(temp[4]),<br>
+ ureg_src(temp[5]));<br>
+ ureg_MUL(shader, temp[7],
ureg_imm1f(shader,<br>
+ 0.101321), a);<br>
+ ureg_MUL(shader, temp[7],
ureg_src(temp[7]),<br>
+ ureg_src(temp[6]));<br>
+ ureg_DIV(shader, temp[7],
ureg_src(temp[7]),<br>
+ ureg_src(temp[0]));<br>
+ ureg_DIV(shader, o_fragment,<br>
+ ureg_src(temp[7]),
ureg_src(temp[0]));<br>
+<br>
+ ureg_LRP(shader, o_fragment,
ureg_src(temp[1]),<br>
+ ureg_imm1f(shader,
1.0f), ureg_src(o_fragment));<br>
+<br>
+ for(i = 0; i < 8; ++i)<br>
+
ureg_release_temporary(shader,
temp[i]);<br>
+}<br>
+<br>
+static void *<br>
+create_frag_shader(struct
vl_lanczos_filter *filter,
unsigned num_offsets,<br>
+ struct
vertex2f *offsets, unsigned a,<br>
+ unsigned
video_width, unsigned
video_height)<br>
+{<br>
+ struct pipe_screen *screen =
filter->pipe->screen;<br>
+ struct ureg_program *shader;<br>
+ struct ureg_src i_vtex, vtex;<br>
+ struct ureg_src sampler;<br>
+ struct ureg_src half_pixel;<br>
+ struct ureg_dst o_fragment;<br>
+ struct ureg_dst *t_array =
MALLOC(sizeof(struct ureg_dst) *
(num_offsets + 2));<br>
+ struct ureg_dst x, t_sum;<br>
+ unsigned i;<br>
+ bool first;<br>
+<br>
+ if
(screen->get_shader_param(<br>
+ screen,
PIPE_SHADER_FRAGMENT,
PIPE_SHADER_CAP_MAX_TEMPS) <
num_offsets + 2) {<br>
+ return NULL;<br>
+ }<br>
+<br>
+ shader =
ureg_create(PIPE_SHADER_FRAGMENT);<br>
+ if (!shader) {<br>
+ return NULL;<br>
+ }<br>
+<br>
+ i_vtex =
ureg_DECL_fs_input(shader,
TGSI_SEMANTIC_GENERIC, VS_O_VTEX,
TGSI_INTERPOLATE_LINEAR);<br>
+ sampler =
ureg_DECL_sampler(shader, 0);<br>
+<br>
+ for (i = 0; i < num_offsets
+ 2; ++i)<br>
+ t_array[i] =
ureg_DECL_temporary(shader);<br>
+ x =
ureg_DECL_temporary(shader);<br>
+<br>
+ half_pixel =
ureg_DECL_constant(shader, 0);<br>
+ o_fragment =
ureg_DECL_output(shader,
TGSI_SEMANTIC_COLOR, 0);<br>
+<br>
+ /*<br>
+ * temp = (i_vtex -
(0.5/dst_size)) * i_size)<br>
+ * x = frac(temp)<br>
+ * vtex = floor(i_vtex)/i_size
+ half_pixel<br>
+ */<br>
</blockquote>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>This seems to be the reason for the
artifacts.</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<div>
<div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
+ ureg_SUB(shader,
ureg_writemask(t_array[0],
TGSI_WRITEMASK_XY),<br>
+ i_vtex, half_pixel);<br>
</blockquote>
</div>
</div>
</blockquote>
<div><br>
</div>
<div>On debugging I found that after
removing this ^^^ instruction the
artifacts are gone.</div>
<div>Not sure why is this happening but
the filter is working fine.</div>
<div><br>
</div>
<div>Any ideas Christian?</div>
<div><br>
</div>
<div>Regards,</div>
<div>Nayan. </div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
<div>
<div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">
+ ureg_MUL(shader,
ureg_writemask(t_array[1],
TGSI_WRITEMASK_XY),<br>
+ ureg_src(t_array[0]),
ureg_imm2f(shader, video_width,
video_height));<br>
+ ureg_FRC(shader,
ureg_writemask(x,
TGSI_WRITEMASK_XY),<br>
+
ureg_src(t_array[1]));<br>
+<br>
+ ureg_FLR(shader,
ureg_writemask(t_array[1],
TGSI_WRITEMASK_XY),<br>
+
ureg_src(t_array[1]));<br>
+ ureg_DIV(shader,
ureg_writemask(t_array[1],
TGSI_WRITEMASK_XY),<br>
+ ureg_src(t_array[1]),
ureg_imm2f(shader, video_width,
video_height));<br>
+ ureg_ADD(shader,
ureg_writemask(t_array[1],
TGSI_WRITEMASK_XY),<br>
+ ureg_src(t_array[1]),
half_pixel);<br>
+ /*<br>
+ * t_array[2..*] = vtex +
offset[0..*]<br>
+ * t_array[2..*] =
tex(t_array[0..*], sampler)<br>
+ * o_fragment = sum(t_array[i]
* lanczos(x - offsets[i].x) *
lanczos(y - offsets[i].y))<br>
+ */<br>
+ vtex = ureg_src(t_array[1]);<br>
+ for (i = 0; i <
num_offsets; ++i) {<br>
+ ureg_ADD(shader,
ureg_writemask(t_array[i + 2],
TGSI_WRITEMASK_XY),<br>
+ vtex,
ureg_imm2f(shader, offsets[i].x,
offsets[i].y));<br>
+ ureg_MOV(shader,
ureg_writemask(t_array[i + 2],
TGSI_WRITEMASK_ZW),<br>
+
ureg_imm1f(shader, 0.0f));<br>
+ }<br>
+<br>
+ for (i = 0; i <
num_offsets; ++i) {<br>
+ ureg_TEX(shader, t_array[i
+ 2], TGSI_TEXTURE_2D,
ureg_src(t_array[i + 2]),
sampler);<br>
+ }<br>
+<br>
+ for(i = 0, first = true; i
< num_offsets; ++i) {<br>
+ if (first) {<br>
+ t_sum = t_array[i + 2];<br>
+ ureg_SUB(shader,
ureg_writemask(t_array[i],
TGSI_WRITEMASK_XY),<br>
+ ureg_src(x),
ureg_imm2f(shader, offsets[i].x *
video_width,<br>
+ offsets[i].y *
video_height));<br>
+
create_frag_shader_lanczos(shader,
ureg_imm1f(shader, (float)(a)),<br>
+
ureg_scalar(ureg_src(t_array[i]),
TGSI_SWIZZLE_X), t_array[i + 1]);<br>
+
create_frag_shader_lanczos(shader,
ureg_imm1f(shader, (float)(a)),<br>
+
ureg_scalar(ureg_src(t_array[i]),
TGSI_SWIZZLE_Y), t_array[i]);<br>
+ ureg_MUL(shader,
t_array[i + 1], ureg_src(t_array[i
+ 1]),<br>
+
ureg_src(t_array[i]));<br>
+ ureg_MUL(shader, t_sum,
ureg_src(t_array[i + 2]),<br>
+
ureg_src(t_array[i + 1]));<br>
+ first = false;<br>
+ } else {<br>
+ ureg_SUB(shader,
ureg_writemask(t_array[i],
TGSI_WRITEMASK_XY),<br>
+ ureg_src(x),
ureg_imm2f(shader, offsets[i].x *
video_width,<br>
+ offsets[i].y *
video_height));<br>
+
create_frag_shader_lanczos(shader,
ureg_imm1f(shader, (float)(a)),<br>
+
ureg_scalar(ureg_src(t_array[i]),
TGSI_SWIZZLE_X), t_array[i + 1]);<br>
+
create_frag_shader_lanczos(shader,
ureg_imm1f(shader, (float)(a)),<br>
+
ureg_scalar(ureg_src(t_array[i]),
TGSI_SWIZZLE_Y), t_array[i]);<br>
+ ureg_MUL(shader,
t_array[i + 1], ureg_src(t_array[i
+ 1]),<br>
+
ureg_src(t_array[i]));<br>
+ ureg_MAD(shader, t_sum,
ureg_src(t_array[i + 2]),<br>
+
ureg_src(t_array[i + 1]),
ureg_src(t_sum));<br>
+ }<br>
+ }<br>
+<br>
+ if (first)<br>
+ ureg_MOV(shader,
o_fragment, ureg_imm1f(shader,
0.0f));<br>
+ else<br>
+ ureg_MOV(shader,
o_fragment, ureg_src(t_sum));<br>
+<br>
+ ureg_release_temporary(shader,
x);<br>
+ ureg_END(shader);<br>
+<br>
+ FREE(t_array);<br>
+ return
ureg_create_shader_and_destroy(shader,
filter->pipe);<br>
+}<br>
+<br>
+bool<br>
+vl_lanczos_filter_init(struct
vl_lanczos_filter *filter, struct
pipe_context *pipe,<br>
+ unsigned
size, unsigned width, unsigned
height)<br>
+{<br>
+ struct pipe_rasterizer_state
rs_state;<br>
+ struct pipe_blend_state blend;<br>
+ struct vertex2f *offsets, v,
sizes;<br>
+ struct pipe_sampler_state
sampler;<br>
+ struct pipe_vertex_element ve;<br>
+ unsigned i, num_offsets = (2 *
size) * (2 * size);<br>
+<br>
+ assert(filter &&
pipe);<br>
+ assert(width &&
height);<br>
+ assert(size);<br>
+<br>
+ memset(filter, 0,
sizeof(*filter));<br>
+ filter->pipe = pipe;<br>
+<br>
+ memset(&rs_state, 0,
sizeof(rs_state));<br>
+ rs_state.half_pixel_center =
true;<br>
+ rs_state.bottom_edge_rule =
true;<br>
+ rs_state.depth_clip = 1;<br>
+ filter->rs_state =
pipe->create_rasterizer_state(pipe,
&rs_state);<br>
+ if (!filter->rs_state)<br>
+ goto error_rs_state;<br>
+<br>
+ memset(&blend, 0, sizeof
blend);<br>
+ blend.rt[0].rgb_func =
PIPE_BLEND_ADD;<br>
+ blend.rt[0].rgb_src_factor =
PIPE_BLENDFACTOR_ONE;<br>
+ blend.rt[0].rgb_dst_factor =
PIPE_BLENDFACTOR_ONE;<br>
+ blend.rt[0].alpha_func =
PIPE_BLEND_ADD;<br>
+ blend.rt[0].alpha_src_factor =
PIPE_BLENDFACTOR_ONE;<br>
+ blend.rt[0].alpha_dst_factor =
PIPE_BLENDFACTOR_ONE;<br>
+ blend.logicop_func =
PIPE_LOGICOP_CLEAR;<br>
+ blend.rt[0].colormask =
PIPE_MASK_RGBA;<br>
+ filter->blend =
pipe->create_blend_state(pipe,
&blend);<br>
+ if (!filter->blend)<br>
+ goto error_blend;<br>
+<br>
+ memset(&sampler, 0,
sizeof(sampler));<br>
+ sampler.wrap_s =
PIPE_TEX_WRAP_CLAMP_TO_EDGE;<br>
+ sampler.wrap_t =
PIPE_TEX_WRAP_CLAMP_TO_EDGE;<br>
+ sampler.wrap_r =
PIPE_TEX_WRAP_CLAMP_TO_EDGE;<br>
+ sampler.min_img_filter =
PIPE_TEX_FILTER_NEAREST;<br>
+ sampler.min_mip_filter =
PIPE_TEX_MIPFILTER_NONE;<br>
+ sampler.mag_img_filter =
PIPE_TEX_FILTER_NEAREST;<br>
+ sampler.compare_mode =
PIPE_TEX_COMPARE_NONE;<br>
+ sampler.compare_func =
PIPE_FUNC_ALWAYS;<br>
+ sampler.normalized_coords = 1;<br>
+ filter->sampler =
pipe->create_sampler_state(pipe,
&sampler);<br>
+ if (!filter->sampler)<br>
+ goto error_sampler;<br>
+<br>
+ filter->quad =
vl_vb_upload_quads(pipe);<br>
+ if(!filter->quad.buffer)<br>
+ goto error_quad;<br>
+<br>
+ memset(&ve, 0,
sizeof(ve));<br>
+ ve.src_offset = 0;<br>
+ ve.instance_divisor = 0;<br>
+ ve.vertex_buffer_index = 0;<br>
+ ve.src_format =
PIPE_FORMAT_R32G32_FLOAT;<br>
+ filter->ves =
pipe->create_vertex_elements_state(pipe,
1, &ve);<br>
+ if (!filter->ves)<br>
+ goto error_ves;<br>
+<br>
+ offsets = MALLOC(sizeof(struct
vertex2f) * num_offsets);<br>
+ if (!offsets)<br>
+ goto error_offsets;<br>
+<br>
+ sizes.x = (float)(size);<br>
+ sizes.y = (float)(size);<br>
+<br>
+ for (v.x = -sizes.x + 1.0f, i
= 0; v.x <= sizes.x; v.x +=
1.0f)<br>
+ for (v.y = -sizes.y + 1.0f;
v.y <= sizes.y; v.y += 1.0f)<br>
+ offsets[i++] = v;<br>
+<br>
+ for (i = 0; i <
num_offsets; ++i) {<br>
+ offsets[i].x /= width;<br>
+ offsets[i].y /= height;<br>
+ }<br>
+<br>
+ filter->vs =
create_vert_shader(filter);<br>
+ if (!filter->vs)<br>
+ goto error_vs;<br>
+<br>
+ filter->fs =
create_frag_shader(filter,
num_offsets, offsets, size, width,
height);<br>
+ if (!filter->fs)<br>
+ goto error_fs;<br>
+<br>
+ FREE(offsets);<br>
+ return true;<br>
+<br>
+error_fs:<br>
+ pipe->delete_vs_state(pipe,
filter->vs);<br>
+<br>
+error_vs:<br>
+ FREE(offsets);<br>
+<br>
+error_offsets:<br>
+
pipe->delete_vertex_elements_state(pipe,
filter->ves);<br>
+<br>
+error_ves:<br>
+
pipe_resource_reference(&filter->quad.buffer,
NULL);<br>
+<br>
+error_quad:<br>
+
pipe->delete_sampler_state(pipe,
filter->sampler);<br>
+<br>
+error_sampler:<br>
+
pipe->delete_blend_state(pipe,
filter->blend);<br>
+<br>
+error_blend:<br>
+
pipe->delete_rasterizer_state(pipe,
filter->rs_state);<br>
+<br>
+error_rs_state:<br>
+ return false;<br>
+}<br>
+<br>
+void<br>
+vl_lanczos_filter_cleanup(struct
vl_lanczos_filter *filter)<br>
+{<br>
+ assert(filter);<br>
+<br>
+
filter->pipe->delete_sampler_state(filter->pipe,
filter->sampler);<br>
+
filter->pipe->delete_blend_state(filter->pipe,
filter->blend);<br>
+
filter->pipe->delete_rasterizer_state(filter->pipe,
filter->rs_state);<br>
+
filter->pipe->delete_vertex_elements_state(filter->pipe,
filter->ves);<br>
+
pipe_resource_reference(&filter->quad.buffer,
NULL);<br>
+<br>
+
filter->pipe->delete_vs_state(filter->pipe,
filter->vs);<br>
+
filter->pipe->delete_fs_state(filter->pipe,
filter->fs);<br>
+}<br>
+<br>
+void<br>
+vl_lanczos_filter_render(struct
vl_lanczos_filter *filter,<br>
+ struct
pipe_sampler_view *src,<br>
+ struct
pipe_surface *dst,<br>
+ struct
u_rect *dst_area,<br>
+ struct
u_rect *dst_clip)<br>
+{<br>
+ struct pipe_viewport_state
viewport;<br>
+ struct pipe_framebuffer_state
fb_state;<br>
+ struct pipe_scissor_state
scissor;<br>
+ union pipe_color_union
clear_color;<br>
+ struct pipe_transfer
*buf_transfer;<br>
+ struct pipe_resource
*surface_size;<br>
+ assert(filter && src
&& dst);<br>
+<br>
+ if (dst_clip) {<br>
+ scissor.minx =
dst_clip->x0;<br>
+ scissor.miny =
dst_clip->y0;<br>
+ scissor.maxx =
dst_clip->x1;<br>
+ scissor.maxy =
dst_clip->y1;<br>
+ } else {<br>
+ scissor.minx = 0;<br>
+ scissor.miny = 0;<br>
+ scissor.maxx =
dst->width;<br>
+ scissor.maxy =
dst->height;<br>
+ }<br>
+<br>
+ clear_color.f[0] =
clear_color.f[1] = 0.0f;<br>
+ clear_color.f[2] =
clear_color.f[3] = 0.0f;<br>
+ surface_size =
pipe_buffer_create<br>
+ (<br>
+ filter->pipe->screen,<br>
+ PIPE_BIND_CONSTANT_BUFFER,<br>
+ PIPE_USAGE_DEFAULT,<br>
+ 2*sizeof(float)<br>
+ );<br>
+<br>
+<br>
+ memset(&viewport, 0,
sizeof(viewport));<br>
+ if(dst_area){<br>
+ viewport.scale[0] =
dst_area->x1 - dst_area->x0;<br>
+ viewport.scale[1] =
dst_area->y1 - dst_area->y0;<br>
+ viewport.translate[0] =
dst_area->x0;<br>
+ viewport.translate[1] =
dst_area->y0;<br>
+ } else {<br>
+ viewport.scale[0] =
dst->width;<br>
+ viewport.scale[1] =
dst->height;<br>
+ }<br>
+ viewport.scale[2] = 1;<br>
+<br>
+ float *ptr =
pipe_buffer_map(filter->pipe,
surface_size,<br>
+
PIPE_TRANSFER_WRITE |
PIPE_TRANSFER_DISCARD_RANGE,<br>
+
&buf_transfer);<br>
+<br>
+ ptr[0] =
0.5f/viewport.scale[0];<br>
+ ptr[1] =
0.5f/viewport.scale[1];<br>
+<br>
+
pipe_buffer_unmap(filter->pipe,
buf_transfer);<br>
+<br>
+ memset(&fb_state, 0,
sizeof(fb_state));<br>
+ fb_state.width =
dst->width;<br>
+ fb_state.height =
dst->height;<br>
+ fb_state.nr_cbufs = 1;<br>
+ fb_state.cbufs[0] = dst;<br>
+<br>
+
filter->pipe->set_scissor_states(filter->pipe,
0, 1, &scissor);<br>
+
filter->pipe->clear_render_target(filter->pipe,
dst, &clear_color,<br>
+
0, 0, dst->width,
dst->height);<br>
+
pipe_set_constant_buffer(filter->pipe,
PIPE_SHADER_FRAGMENT, 0,
surface_size);<br>
+
filter->pipe->bind_rasterizer_state(filter->pipe,
filter->rs_state);<br>
+
filter->pipe->bind_blend_state(filter->pipe,
filter->blend);<br>
+
filter->pipe->bind_sampler_states(filter->pipe,
PIPE_SHADER_FRAGMENT,<br>
+
0, 1, &filter->sampler);<br>
+
filter->pipe->set_sampler_views(filter->pipe,
PIPE_SHADER_FRAGMENT,<br>
+
0, 1, &src);<br>
+
filter->pipe->bind_vs_state(filter->pipe,
filter->vs);<br>
+
filter->pipe->bind_fs_state(filter->pipe,
filter->fs);<br>
+
filter->pipe->set_framebuffer_state(filter->pipe,
&fb_state);<br>
+
filter->pipe->set_viewport_states(filter->pipe,
0, 1, &viewport);<br>
+
filter->pipe->set_vertex_buffers(filter->pipe,
0, 1, &filter->quad);<br>
+
filter->pipe->bind_vertex_elements_state(filter->pipe,
filter->ves);<br>
+<br>
+
util_draw_arrays(filter->pipe,
PIPE_PRIM_QUADS, 0, 4);<br>
+}<br>
diff --git
a/src/gallium/auxiliary/vl/vl_lanczos_filter.h
b/src/gallium/auxiliary/vl/vl_lanczos_filter.h<br>
new file mode 100644<br>
index 0000000..cb469aa<br>
--- /dev/null<br>
+++
b/src/gallium/auxiliary/vl/vl_lanczos_filter.h<br>
@@ -0,0 +1,63 @@<br>
+/**************************************************************************<br>
+ *<br>
+ * Copyright 2016 Nayan Deshmukh.<br>
+ * All Rights Reserved.<br>
+ *<br>
+ * Permission is hereby granted,
free of charge, to any person
obtaining a<br>
+ * copy of this software and
associated documentation files
(the<br>
+ * "Software"), to deal in the
Software without restriction,
including<br>
+ * without limitation the rights
to use, copy, modify, merge,
publish,<br>
+ * distribute, sub license,
and/or sell copies of the
Software, and to<br>
+ * permit persons to whom the
Software is furnished to do so,
subject to<br>
+ * the following conditions:<br>
+ *<br>
+ * The above copyright notice and
this permission notice (including
the<br>
+ * next paragraph) shall be
included in all copies or
substantial portions<br>
+ * of the Software.<br>
+ *<br>
+ * THE SOFTWARE IS PROVIDED "AS
IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS<br>
+ * OR IMPLIED, INCLUDING BUT NOT
LIMITED TO THE WARRANTIES OF<br>
+ * MERCHANTABILITY, FITNESS FOR A
PARTICULAR PURPOSE AND
NON-INFRINGEMENT.<br>
+ * IN NO EVENT SHALL VMWARE
AND/OR ITS SUPPLIERS BE LIABLE FOR<br>
+ * ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF
CONTRACT,<br>
+ * TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH
THE<br>
+ * SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.<br>
+ *<br>
+
**************************************************************************/<br>
+<br>
+/* implementation of lanczos
interpolation filter */<br>
+<br>
+#ifndef vl_lanczos_filter_h<br>
+#define vl_lanczos_filter_h<br>
+<br>
+#include "pipe/p_state.h"<br>
+<br>
+struct vl_lanczos_filter<br>
+{<br>
+ struct pipe_context *pipe;<br>
+ struct pipe_vertex_buffer
quad;<br>
+<br>
+ void *rs_state;<br>
+ void *blend;<br>
+ void *sampler;<br>
+ void *ves;<br>
+ void *vs, *fs;<br>
+};<br>
+<br>
+bool<br>
+vl_lanczos_filter_init(struct
vl_lanczos_filter *filter, struct
pipe_context *pipe,<br>
+ unsigned
size, unsigned width, unsigned
height);<br>
+<br>
+void<br>
+vl_lanczos_filter_cleanup(struct
vl_lanczos_filter *filter);<br>
+<br>
+<br>
+void<br>
+vl_lanczos_filter_render(struct
vl_lanczos_filter *filter,<br>
+ struct
pipe_sampler_view *src,<br>
+ struct
pipe_surface *dst,<br>
+ struct
u_rect *dst_area,<br>
+ struct
u_rect *dst_clip);<br>
+<br>
+<br>
+#endif /* vl_lanczos_filter_h */<br>
</blockquote>
<br>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</blockquote>
<p><br>
</p>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</div>
</blockquote>
<p><br>
</p>
</div></div></div>
</blockquote></div><br></div>