X-Git-Url: https://git.nihav.org/?a=blobdiff_plain;f=nihav-realmedia%2Fsrc%2Fcodecs%2Frv60dsp.rs;h=8a030e962bd32d1ba4e9fbf5fddede0e1e19092d;hb=0091a5084a0eb195f5f3a7d6924d8885f1c769d3;hp=3c88f39cd88ee78982a61747f1864fb86e3a2aa9;hpb=e07387c7f125550a41faf36a509b252cf71d7f9a;p=nihav.git

diff --git a/nihav-realmedia/src/codecs/rv60dsp.rs b/nihav-realmedia/src/codecs/rv60dsp.rs
index 3c88f39..8a030e9 100644
--- a/nihav-realmedia/src/codecs/rv60dsp.rs
+++ b/nihav-realmedia/src/codecs/rv60dsp.rs
@@ -1,6 +1,6 @@
 use nihav_core::frame::{NAVideoBuffer, NASimpleVideoFrame};
-use nihav_core::codecs::MV;
-use nihav_core::codecs::blockdsp::edge_emu;
+use nihav_codec_support::codecs::MV;
+use nihav_codec_support::codecs::blockdsp::edge_emu;
 
 fn clip8(val: i16) -> u8 { val.min(255).max(0) as u8 }
 
@@ -58,10 +58,11 @@ macro_rules! filter_row {
         });
 }
 
+#[allow(clippy::cognitive_complexity)]
 fn luma_mc(dst: &mut [u8], mut didx: usize, dstride: usize, src: &[u8], mut sidx: usize, sstride: usize, w: usize, h: usize, cx: usize, cy: usize) {
     if (cx == 0) && (cy == 0) {
         for _ in 0..h {
-            for x in 0..w { dst[didx + x] = src[sidx + x]; }
+            dst[didx..][..w].copy_from_slice(&src[sidx..][..w]);
             didx += dstride;
             sidx += sstride;
         }
@@ -102,17 +103,19 @@ fn luma_mc(dst: &mut [u8], mut didx: usize, dstride: usize, src: &[u8], mut sidx
 fn chroma_mc(dst: &mut [u8], mut didx: usize, dstride: usize, src: &[u8], mut sidx: usize, sstride: usize, w: usize, h: usize, x: usize, y: usize) {
     if (x == 0) && (y == 0) {
         for _ in 0..h {
-            for x in 0..w { dst[didx + x] = src[sidx + x]; }
+            dst[didx..][..w].copy_from_slice(&src[sidx..][..w]);
             didx += dstride;
             sidx += sstride;
         }
         return;
     }
     if (x > 0) && (y > 0) {
-        let a = ((4 - x) * (4 - y)) as u16;
-        let b = ((    x) * (4 - y)) as u16;
-        let c = ((4 - x) * (    y)) as u16;
-        let d = ((    x) * (    y)) as u16;
+        // 3,3 case is the same as 3,2 for some reason
+        let ymod = if (x == 3) && (y == 3) { 2 } else { y };
+        let a = ((4 - x) * (4 - ymod)) as u16;
+        let b = ((    x) * (4 - ymod)) as u16;
+        let c = ((4 - x) * (    ymod)) as u16;
+        let d = ((    x) * (    ymod)) as u16;
         for _ in 0..h {
             for x in 0..w {
                 dst[didx + x] = ((a * (src[sidx + x] as u16)
@@ -151,7 +154,7 @@ macro_rules! diff{
         )
 }
 macro_rules! strength{
-    ($el: expr, $lim: expr) => (if $el < $lim { 3 } else { 1 })
+    ($el: expr, $lim: expr) => (if $el.abs() < $lim { 3 } else { 1 })
 }
 fn clip_symm(val: i16, lim: i16) -> i16 { val.max(-lim).min(lim) }
 
@@ -167,7 +170,7 @@ fn filter_luma_edge(dst: &mut [u8], mut offset: usize, step: usize, stride: usiz
     let str_q = strength!(diff_p1p0[0] + diff_p1p0[1] + diff_p1p0[2] + diff_p1p0[3], lim2);
     if str_p + str_q > 2 {
         let msum = ((mode1 + mode2 + str_q + str_p) >> 1) as i16;
-        let (maxprod, weak) = if (str_q == 1) || (str_p == 1) { (512, true) } else { (384, false) }; 
+        let (maxprod, weak) = if (str_q == 1) || (str_p == 1) { (512, true) } else { (384, false) };
         for y in 0..4 {
             let diff_p0q0 = diff!(dst, offset, offset - step);
             if (diff_p0q0 != 0) && (lim1 * diff_p0q0.abs() < maxprod) {
@@ -181,7 +184,7 @@ fn filter_luma_edge(dst: &mut [u8], mut offset: usize, step: usize, stride: usiz
                     };
                 dst[offset - step] = clip8((dst[offset - step] as i16) + delta);
                 dst[offset]        = clip8((dst[offset]        as i16) - delta);
-                if (str_q != 1) && (diff_q1q2.abs() <= (lim1 >> 2)) {
+                if (str_q != 1) && (diff_q1q2.abs() <= (lim2 >> 2)) {
                     let diff = (diff_q1q0[y] + diff_q1q2 - delta) >> 1;
                     let delta_q1 = if weak {
                             clip_symm(diff, (mode1 >> 1) as i16)
@@ -190,7 +193,7 @@ fn filter_luma_edge(dst: &mut [u8], mut offset: usize, step: usize, stride: usiz
                         };
                     dst[offset - 2 * step] = clip8((dst[offset - 2 * step] as i16) - delta_q1);
                 }
-                if (str_p != 1) && (diff_p1p2.abs() <= (lim1 >> 2)) {
+                if (str_p != 1) && (diff_p1p2.abs() <= (lim2 >> 2)) {
                     let diff = (diff_p1p0[y] + diff_p1p2 + delta) >> 1;
                     let delta_p1 = if weak {
                             clip_symm(diff, (mode2 >> 1) as i16)
@@ -408,7 +411,7 @@ impl RV60DSP {
         blk[off +  0 * step] = ((tm0 + tt0 + 64) >> 7) as i16;
         blk[off +  1 * step] = ((tm4 + tt1 + 64) >> 7) as i16;
         blk[off +  2 * step] = ((tm6 + tt2 + 64) >> 7) as i16;
-        blk[off +  3 * step] = ((tm4 + tt3 + 64) >> 7) as i16;
+        blk[off +  3 * step] = ((tm2 + tt3 + 64) >> 7) as i16;
         blk[off +  4 * step] = ((tm3 + tt4 + 64) >> 7) as i16;
         blk[off +  5 * step] = ((tm7 + tt5 + 64) >> 7) as i16;
         blk[off +  6 * step] = ((tm5 + tt6 + 64) >> 7) as i16;
@@ -491,7 +494,7 @@ impl RV60DSP {
                 luma_mc(dst, doffset, dstride, src, soffset, sstride, w, h, cx, cy);
             } else {
                 let mut ebuf: [u8; 70*70] = [0; 70*70];
-                edge_emu(prev_frame, (x as isize) + (dx as isize) - 2, (y as isize) + (dy as isize) - 2, w+5, h+5, &mut ebuf, 70, 0);
+                edge_emu(prev_frame, (x as isize) + (dx as isize) - 2, (y as isize) + (dy as isize) - 2, w+5, h+5, &mut ebuf, 70, 0, 4);
                 luma_mc(dst, doffset, dstride, &ebuf, 70*2 + 2, 70, w, h, cx, cy);
             }
         }
@@ -519,7 +522,7 @@ impl RV60DSP {
                 chroma_mc(frame.data, doffset, dstride, src, soffset, sstride, cw, ch, cx, cy);
             } else {
                 let mut ebuf: [u8; 40*40] = [0; 40*40];
-                edge_emu(prev_frame, ((x >> 1) as isize) + (dx as isize), ((y >> 1) as isize) + (dy as isize), cw+1, ch+1, &mut ebuf, 40, comp);
+                edge_emu(prev_frame, ((x >> 1) as isize) + (dx as isize), ((y >> 1) as isize) + (dy as isize), cw+1, ch+1, &mut ebuf, 40, comp, 3);
                 chroma_mc(frame.data, doffset, dstride, &ebuf, 0, 40, cw, ch, cx, cy);
             }
         }
@@ -542,7 +545,7 @@ impl RV60DSP {
             filter_luma_edge(frame.data, offset, 1, stride, mode_l, mode_r, lim1, lim2);
         }
         if ((str_l | str_r) >= 2) && deblock_chroma {
-            for comp in 1..2 {
+            for comp in 1..3 {
                 let stride = frame.stride[comp];
                 let offset = frame.offset[comp] + (xpos >> 1) + (ypos >> 1) * stride;
                 filter_chroma_edge(frame.data, offset, 1, stride, mode_l, mode_r, lim1, lim2);
@@ -567,7 +570,7 @@ impl RV60DSP {
             filter_luma_edge(frame.data, offset, stride, 1, mode_t, mode_d, lim1, lim2);
         }
         if ((str_t | str_d) >= 2) && deblock_chroma {
-            for comp in 1..2 {
+            for comp in 1..3 {
                 let stride = frame.stride[comp];
                 let offset = frame.offset[comp] + (xpos >> 1) + (ypos >> 1) * stride;
                 filter_chroma_edge(frame.data, offset, stride, 1, mode_t, mode_d, lim1, lim2);
@@ -578,46 +581,46 @@ impl RV60DSP {
                   xpos: usize, ypos: usize, top_str: &[u8], left_str: &[u8], dblkpos: usize) {
         if xpos > 0 {
             if ypos > 0 {
-                let str_l = left_str[dblkpos - dparams.dblkstride];
-                let str_r = left_str[dblkpos];
-                if (str_l | str_r) != 0 {
+                let str_l = left_str[dblkpos - dparams.dblkstride - 1];
+                let str_r = left_str[dblkpos - dparams.dblkstride];
+                if ((str_l | str_r) & 3) != 0 {
                     self.deblock_edge4_ver(frame, xpos, ypos - 4, str_l, str_r, dparams.deblock_chroma);
                 }
             }
             {
-                let str_l = left_str[dblkpos];
-                let str_r = left_str[dblkpos + dparams.dblkstride];
-                if (str_l | str_r) != 0 {
+                let str_l = left_str[dblkpos - 1];
+                let str_r = left_str[dblkpos];
+                if ((str_l | str_r) & 3) != 0 {
                     self.deblock_edge4_ver(frame, xpos, ypos + 0, str_l, str_r, dparams.deblock_chroma);
                 }
             }
-            if ypos + 4 >= dparams.height {
-                let str_l = left_str[dblkpos + dparams.dblkstride];
-                let str_r = left_str[dblkpos + dparams.dblkstride * 2];
-                if (str_l | str_r) != 0 {
+            if ypos + 8 >= dparams.height {
+                let str_l = left_str[dblkpos + dparams.dblkstride - 1];
+                let str_r = left_str[dblkpos + dparams.dblkstride];
+                if ((str_l | str_r) & 3) != 0 {
                     self.deblock_edge4_ver(frame, xpos, ypos + 4, str_l, str_r, dparams.deblock_chroma);
                 }
             }
         }
         if ypos > 0 {
             if xpos > 0 {
-                let str_t = top_str[dblkpos - 1];
-                let str_d = top_str[dblkpos];
-                if (str_t | str_d) != 0 {
+                let str_t = top_str[dblkpos - dparams.dblkstride - 1];
+                let str_d = top_str[dblkpos - 1];
+                if ((str_t | str_d) & 3) != 0 {
                     self.deblock_edge4_hor(frame, xpos - 4, ypos, str_t, str_d, dparams.deblock_chroma);
                 }
             }
             {
-                let str_t = top_str[dblkpos];
-                let str_d = top_str[dblkpos + 1];
-                if (str_t | str_d) != 0 {
+                let str_t = top_str[dblkpos - dparams.dblkstride];
+                let str_d = top_str[dblkpos];
+                if ((str_t | str_d) & 3) != 0 {
                     self.deblock_edge4_hor(frame, xpos + 0, ypos, str_t, str_d, dparams.deblock_chroma);
                 }
             }
-            if xpos + 4 >= dparams.width {
-                let str_t = top_str[dblkpos + 1];
-                let str_d = top_str[dblkpos + 2];
-                if (str_t | str_d) != 0 {
+            if xpos + 8 >= dparams.width {
+                let str_t = top_str[dblkpos - dparams.dblkstride + 1];
+                let str_d = top_str[dblkpos + 1];
+                if ((str_t | str_d) & 3) != 0 {
                     self.deblock_edge4_hor(frame, xpos + 4, ypos, str_t, str_d, dparams.deblock_chroma);
                 }
             }
@@ -757,9 +760,7 @@ impl IntraPredContext {
             let off = ((sum >> 5) + 32) as usize;
             let frac = (sum & 0x1F) as u16;
             if frac == 0 {
-                for x in 0..size {
-                    dst[doff + x] = src[off + x];
-                }
+                dst[doff..][..size].copy_from_slice(&src[off..][..size]);
             } else {
                 for x in 0..size {
                     let a = src[off + x + 0] as u16;
@@ -785,6 +786,7 @@ impl IntraPredContext {
             sum += diff;
         }
     }
+    #[allow(clippy::cognitive_complexity)]
     pub fn pred_angle(&self, dst: &mut [u8], mut doff: usize, dstride: usize, size: usize, angle: usize, filter: bool) {
         let mut filtered1: [u8; 96] = [0; 96];
         let mut filtered2: [u8; 96] = [0; 96];
@@ -880,9 +882,7 @@ impl IntraPredContext {
                 Self::filter_bilin32(&mut filtered1[32..], self.t[1], self.t[33], 32);
             }
             for _ in 0..size {
-                for x in 0..size {
-                    dst[doff + x] = filtered1[32 + x];
-                }
+                dst[doff..][..size].copy_from_slice(&filtered1[32..][..size]);
                 doff += dstride;
             }
             if filter {