diff --git a/effects/scaler/Makefile b/effects/scaler/Makefile
new file mode 100644
index 00000000..86a04c5a
--- /dev/null
+++ b/effects/scaler/Makefile
@@ -0,0 +1,8 @@
+TOPDIR := $(realpath ../..)
+
+CLEAN-FILES := data/stars-die.c
+SUBDIRS := scalertab
+
+PNG2C.stars-die := --bitmap image,320x256x5,+interleaved --palette image_pal,32
+
+include $(TOPDIR)/build/effect.mk
diff --git a/effects/scaler/data/stars-die.png b/effects/scaler/data/stars-die.png
new file mode 100644
index 00000000..0faf9936
--- /dev/null
+++ b/effects/scaler/data/stars-die.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7aecfc44106b2d8168984b948dc78f6be1509d3c6f1db368fac17b2686fb0863
+size 24313
diff --git a/effects/scaler/scaler.c b/effects/scaler/scaler.c
new file mode 100644
index 00000000..aee139f6
--- /dev/null
+++ b/effects/scaler/scaler.c
@@ -0,0 +1,148 @@
+#include "effect.h"
+#include "custom.h"
+#include "copper.h"
+#include "bitmap.h"
+#include "palette.h"
+#include "profiler.h"
+
+#include "data/stars-die.c"
+
+#define LINES 256
+#define DEPTH 5
+
+static CopListT *cp[2];
+static short active = 0;
+
+static void VerticalScalerForward(CopListT *cp, short ys, short height) {
+  short rowmod = image.bytesPerRow * image.depth;
+  int dy = (LINES << 16) / height;
+  short n = (short)(dy >> 16) * (short)image.depth;
+  short mod = (short)image.bytesPerRow * (short)(n - 1);
+  int y = 0;
+  short i;
+
+  for (i = 1; i < height; i++) {
+    short _mod = mod;
+    int ny = y + dy;
+    if ((u_short)ny < (u_short)y)
+      _mod += rowmod;
+    CopWaitSafe(cp, Y(ys + i), X(0));
+    CopMove16(cp, bpl1mod, _mod);
+    CopMove16(cp, bpl2mod, _mod);
+    y = ny;
+  }
+}
+
+static void CopSetupBitplanesReverse(CopListT *list, CopInsT **bplptr,
+                                     const BitmapT *bitmap, u_short depth)
+{
+  short bytesPerLine = bitmap->bytesPerRow;
+  int start;
+
+  if (bitmap->flags & BM_INTERLEAVED)
+    bytesPerLine *= (short)bitmap->depth;
+
+  start = bytesPerLine * (short)(bitmap->height - 1);
+
+  {
+    void **planes = bitmap->planes;
+    short n = depth - 1;
+    short reg = CSREG(bplpt);
+
+    do {
+      CopInsT *ins = CopMoveLong(list, reg, (int)(*planes++) + start);
+
+      if (bplptr)
+        *bplptr++ = ins;
+
+      reg += 4;
+    } while (--n != -1);
+  }
+
+  {
+    short modulo;
+
+    if (bitmap->flags & BM_INTERLEAVED)
+      modulo = (short)bitmap->bytesPerRow * (short)(depth + 1);
+    else
+      modulo = 2 * bitmap->bytesPerRow;
+
+    CopMove16(list, bpl1mod, -modulo);
+    CopMove16(list, bpl2mod, -modulo);
+  }
+}
+
+static void VerticalScalerReverse(CopListT *cp, short ys, short height) {
+  short rowmod = image.bytesPerRow * image.depth;
+  int dy = (LINES << 16) / height;
+  short n = (short)(dy >> 16) * (short)image.depth;
+  short mod = (short)image.bytesPerRow * (short)(n + 1);
+  int y = 0;
+  short i;
+
+  for (i = 1; i < height; i++) {
+    short _mod = mod;
+    int ny = y + dy;
+    if ((u_short)ny < (u_short)y)
+      _mod += rowmod;
+    CopWaitSafe(cp, Y(ys + i), X(0));
+    CopMove16(cp, bpl1mod, -_mod);
+    CopMove16(cp, bpl2mod, -_mod);
+    y = ny;
+  }
+}
+
+static void MakeCopperList(CopListT *cp, short height) {
+  short ys = (LINES - abs(height)) / 2;
+
+  CopInit(cp);
+  CopSetupDisplayWindow(cp, MODE_LORES, X(0), Y(ys), image.width, abs(height));
+  if (height > 0) {
+    CopSetupBitplanes(cp, NULL, &image, image.depth);
+    VerticalScalerForward(cp, ys, height);
+  } else if (height < 0) {
+    CopSetupBitplanesReverse(cp, NULL, &image, image.depth);
+    VerticalScalerReverse(cp, ys, -height);
+  }
+  CopEnd(cp);
+}
+
+static void Init(void) {
+  LoadPalette(&image_pal, 0);
+  SetupPlayfield(MODE_LORES, image.depth,
+                 X(0), Y(0), image.width, image.height);
+
+  cp[0] = NewCopList(40 + LINES * 3);
+  cp[1] = NewCopList(40 + LINES * 3);
+  MakeCopperList(cp[active], LINES);
+  CopListActivate(cp[active]);
+
+  EnableDMA(DMAF_RASTER);
+}
+
+static void Kill(void) {
+  DisableDMA(DMAF_COPPER | DMAF_RASTER);
+  DeleteCopList(cp[0]);
+  DeleteCopList(cp[1]);
+}
+
+PROFILE(Scaler);
+
+static void Render(void) {
+  static short val = LINES, dir = -1;
+
+  ProfilerStart(Scaler);
+  MakeCopperList(cp[active], val);
+  ProfilerStop(Scaler);
+
+  CopListRun(cp[active]);
+  TaskWaitVBlank();
+
+  val += dir;
+  if (abs(val) == LINES)
+    dir = -dir;
+
+  active ^= 1;
+}
+
+EFFECT(scaler, NULL, NULL, Init, Kill, Render);
diff --git a/effects/scaler/scalertab/.gitignore b/effects/scaler/scalertab/.gitignore
new file mode 100644
index 00000000..d675eb8e
--- /dev/null
+++ b/effects/scaler/scalertab/.gitignore
@@ -0,0 +1 @@
+scalertab
diff --git a/effects/scaler/scalertab/Makefile b/effects/scaler/scalertab/Makefile
new file mode 100644
index 00000000..339e6e1c
--- /dev/null
+++ b/effects/scaler/scalertab/Makefile
@@ -0,0 +1,3 @@
+TOPDIR := $(realpath ../../..)
+
+include $(TOPDIR)/build/go.mk
diff --git a/effects/scaler/scalertab/main.go b/effects/scaler/scalertab/main.go
new file mode 100644
index 00000000..969ef43a
--- /dev/null
+++ b/effects/scaler/scalertab/main.go
@@ -0,0 +1,75 @@
+package main
+
+import (
+	"fmt"
+	"math"
+	"sort"
+)
+
+type Row struct {
+	err      float64 // average error
+	columns  []int   // current set of columns
+	removed  int     // which column was removed from working image
+	original int     // the column from original picture that was removed
+}
+
+func estimate(S []int, n int, N int) []Row {
+	if n == 1 {
+		R := make([]Row, 1)
+		R[0] = Row{0.0, []int{S[0]}, 1, S[1]}
+		return R
+	}
+
+	di := float64(N-1) / float64(n-1)
+	I := make([]float64, n)
+	for i := 0; i < n; i++ {
+		I[i] = float64(i) * di
+	}
+
+	rows := make([]Row, n+1)
+
+	for i := 0; i <= n; i++ {
+		C := make([]int, n)
+		sum := 0.0
+		for j := 0; j < n; j++ {
+			if j >= i {
+				C[j] = S[j+1]
+			} else {
+				C[j] = S[j]
+			}
+			delta := I[j] - float64(C[j])
+			sum += delta * delta
+		}
+		rows[i] = Row{math.Sqrt(sum) / float64(n), C, i, S[i]}
+	}
+
+	sort.Slice(rows, func(i, j int) bool {
+		return rows[i].err < rows[j].err
+	})
+
+	R := estimate(rows[0].columns, n-1, N)
+	for _, row := range rows[1:1] {
+		C := estimate(row.columns, n-1, N)
+		if R[len(R)-1].err > C[len(C)-1].err {
+			R = C
+		}
+	}
+
+	return append(R, rows[0])
+}
+
+func main() {
+	N := 40
+	S := make([]int, N)
+	for i := 0; i < N; i++ {
+		S[i] = i
+	}
+	R := estimate(S, N-1, N)
+	R = append(R, Row{0.0, S, -1, -1})
+	err := 0.0
+	for i, row := range R {
+		fmt.Printf("%3d: [%d] = %d\n", i+1, row.removed, row.original)
+		err += row.err
+	}
+	fmt.Printf("\naverage error: %.3f\n", err/float64(N-1))
+}
diff --git a/lib/libgfx/CopSetupBitplanes.c b/lib/libgfx/CopSetupBitplanes.c
index eafdcb6c..b6b67d60 100644
--- a/lib/libgfx/CopSetupBitplanes.c
+++ b/lib/libgfx/CopSetupBitplanes.c
@@ -6,13 +6,15 @@ void CopSetupBitplanes(CopListT *list, CopInsT **bplptr,
   {
     void **planes = bitmap->planes;
     short n = depth - 1;
-    short i = 0;
+    short reg = CSREG(bplpt);
 
     do {
-      CopInsT *ins = CopMove32(list, bplpt[i++], *planes++);
+      CopInsT *ins = CopMoveLong(list, reg, (int)(*planes++));
 
       if (bplptr)
         *bplptr++ = ins;
+
+      reg += 4;
     } while (--n != -1);
   }