diff --git a/apps/com.example.camtest/assets/camtest.py b/apps/com.example.camtest/assets/camtest.py
index 18dd1148..9cb4493b 100644
--- a/apps/com.example.camtest/assets/camtest.py
+++ b/apps/com.example.camtest/assets/camtest.py
@@ -20,6 +20,32 @@
 #define CAM_LEDC_CHANNEL    LEDC_CHANNEL_0
 
 
+# OV5640 I2C address
+#sensor_addr = 0x3C
+#reg_addr = 0x4300
+#reg_value = 0x4F  # RGB565 (bit[6]=1) + sequence 0xF (bits[3:0]=0xF)
+#from machine import I2C, Pin
+#i2c = I2C(scl=Pin(16), sda=Pin(21))
+#i2c.readfrom_mem(0x3C, 0x4300, 1)
+#i2c.writeto_mem(sensor_addr, reg_addr, bytes([reg_value]))
+
+
+subwindow.clean()
+canary = lv.obj(subwindow)
+canary.add_flag(lv.obj.FLAG.HIDDEN)
+
+width = 160
+height = 120
+#width = 120
+#height = 160
+
+image = lv.image(subwindow)
+image.align(lv.ALIGN.CENTER, 0, 0)
+#image.set_size(width, height)
+#image.set_size(height, width)
+#image.set_rotation(900)
+
+
 from camera import Camera, GrabMode, PixelFormat, FrameSize, GainCeiling
 
 cam = Camera(
@@ -33,47 +59,49 @@ cam = Camera(
     xclk_freq=20000000,
     powerdown_pin=-1,
     reset_pin=-1,
-    #pixel_format=PixelFormat.RGB565,frame_size=FrameSize.QQVGA,grab_mode=GrabMode.LATEST # 160x120, FrameSize.QVGA = 320x240
-    #pixel_format=PixelFormat.JPEG,frame_size=FrameSize.QVGA,grab_mode=GrabMode.LATEST
+    pixel_format=PixelFormat.RGB565,frame_size=FrameSize.QQVGA,grab_mode=GrabMode.LATEST # 160x120, FrameSize.QVGA = 320x240
 )
 
-cam.init()
+#cam.init() done default (see above)
+
+cam.set_vflip(True)
+
+# LVGL’s expected RGB565 format is likely little-endian: GGGBBBBB RRRRRGGG, since swapping bytes fixed the issue.
+# The sensor’s default sequence is 0x0 ({b[4:0],g[5:3]},{g[2:0],r[4:0]}), which is big-endian-like (BGR in the first byte, RG in the second).
+# After swapping bytes, you get {g[2:0],r[4:0]},{b[4:0],g[5:3]}, which matches LVGL’s expectation.
+# From the datasheet, the sequence {g[2:0],r[4:0]},{b[4:0],g[5:3]} corresponds to 0xF.
+# This suggests setting bits[3:0] to 0xF should produce the correct order without manual swapping.
+
+# default writes:
+# 0x1: {r[4:0],g[5:3]},{g[2:0],b[4:0]}   so that's RRRRR GGGGGG BBBBB
+
+def try_capture():
+    if cam.frame_available():
+        img = bytes(cam.capture())
+        cam.free_buffer()
+        # Swap bytes for each 16-bit pixel
+        #img_swapped = bytearray(len(img))
+        #for i in range(0, len(img), 2):
+        #    img_swapped[i] = img[i+1]    # Swap high and low bytes
+        #    img_swapped[i+1] = img[i]
+        image_dsc = lv.image_dsc_t({
+            "header": { "magic": lv.IMAGE_HEADER_MAGIC, "w": width, "h": height, "stride": width * 2, "cf": lv.COLOR_FORMAT.RGB565 },
+            'data_size': len(img),
+            'data': img
+        })
+        image.set_src(image_dsc)
+
+try_capture()
+
+while canary.is_valid():
+    try_capture()
+
+
+#pixel_format=PixelFormat.JPEG,frame_size=FrameSize.QVGA,grab_mode=GrabMode.LATEST
 
 #cam.reconfigure(pixel_format=PixelFormat.RAW,frame_size=FrameSize.QVGA,grab_mode=GrabMode.LATEST, fb_count=2)
-#cam.reconfigure(pixel_format=PixelFormat.YUV420,frame_size=FrameSize.QVGA,grab_mode=GrabMode.LATEST, fb_count=2)
-
-#img = cam.capture()
-
-def memoryview_to_hex_spaced(mv: memoryview) -> str:
-    """Convert the first 50 bytes of a memoryview to a spaced hex string."""
-    sliced = mv[:50]
-    return ' '.join('{:02x}'.format(b & 0xFF) for b in sliced)
+#cam.reconfigure(pixel_format=PixelFormat.YUV420,frame_size=FrameSize.QQVGA,grab_mode=GrabMode.LATEST, fb_count=2)
+#cam.reconfigure(pixel_format=PixelFormat.RGB565,frame_size=FrameSize.QQVGA,grab_mode=GrabMode.LATEST, fb_count=2)
 
 #memoryview_to_hex_spaced(img)
 
-
-def capture():
-    width = 160
-    height = 120
-    image = lv.image(subwindow)
-    image_data = cam.capture()
-    memoryview_to_hex_spaced(image_data)
-    image_dsc = lv.image_dsc_t({
-        "header": {
-            "magic": lv.IMAGE_HEADER_MAGIC,
-            "w": width,
-            "h": height,
-            "stride": width * 2,
-            "cf": lv.COLOR_FORMAT.RGB565
-            },
-        'data_size': len(image_data),
-        'data': image_data
-    })
-    image.set_src(image_dsc)
-    image.align(lv.ALIGN.TOP_MID, 0, 0)
-    image.set_size(width, height)
-
-#for _ in range(10):
-#    capture()
-
-capture()