diff --git a/components/esp_driver_dma/src/async_color_convert_dma2d.c b/components/esp_driver_dma/src/async_color_convert_dma2d.c
index 720ffa7656d..976a962401e 100644
--- a/components/esp_driver_dma/src/async_color_convert_dma2d.c
+++ b/components/esp_driver_dma/src/async_color_convert_dma2d.c
@@ -35,7 +35,8 @@ struct async_color_convert_transaction {
     dma2d_trans_config_t dma2d_trans_config;  // Per-request DMA2D transaction configuration
 
     async_color_convert_request_t request;    // Cached user request used to build DMA2D transaction
-    dma2d_csc_config_t tx_csc;                // Cached DMA2D CSC configuration resolved in task context
+    dma2d_csc_config_t tx_csc;                // Cached DMA2D TX CSC configuration resolved in task context
+    dma2d_csc_config_t rx_csc;                // Cached DMA2D RX CSC configuration resolved in task context
     async_color_convert_isr_cb_t cb_isr;      // User ISR callback for this request
     void *cb_args;                            // User callback argument
     async_color_convert_dma2d_context_t *ctx; // Back pointer to parent context
@@ -62,34 +63,65 @@ static esp_err_t async_color_convert_dma2d_convert(async_color_convert_context_t
                                                    async_color_convert_isr_cb_t cb_isr,
                                                    void *cb_args);
 
-static bool try_convert_request_to_dma2d_csc(const async_color_convert_request_t *request,
-                                             dma2d_csc_config_t *out_tx_csc)
+static bool is_rgb24_or_bgr24_fourcc(esp_color_fourcc_t fourcc)
+{
+    return fourcc == ESP_COLOR_FOURCC_BGR24 || fourcc == ESP_COLOR_FOURCC_RGB24;
+}
+
+static dma2d_csc_config_t default_tx_csc_config(void)
+{
+    return (dma2d_csc_config_t) {
+        .tx_csc_option = DMA2D_CSC_TX_NONE,
+        .pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0,
+        .post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0,
+    };
+}
+
+static dma2d_csc_config_t default_rx_csc_config(void)
+{
+    return (dma2d_csc_config_t) {
+        .rx_csc_option = DMA2D_CSC_RX_NONE,
+        .pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0,
+        .post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0,
+    };
+}
+
+static bool resolve_dma2d_csc_configs(const async_color_convert_request_t *request,
+                                      dma2d_csc_config_t *out_tx_csc,
+                                      dma2d_csc_config_t *out_rx_csc)
 {
     esp_color_fourcc_t src_fourcc = request->src_color_format;
     esp_color_fourcc_t dst_fourcc = request->dst_color_format;
+    bool src_is_rgb24_or_bgr24 = is_rgb24_or_bgr24_fourcc(src_fourcc);
+    bool dst_is_rgb24_or_bgr24 = is_rgb24_or_bgr24_fourcc(dst_fourcc);
+
+    *out_tx_csc = default_tx_csc_config();
+    *out_rx_csc = default_rx_csc_config();
 
     if (src_fourcc == dst_fourcc) {
-        out_tx_csc->tx_csc_option = DMA2D_CSC_TX_NONE;
-        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
-        out_tx_csc->post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
+        return true;
+    }
+
+    if (src_is_rgb24_or_bgr24 && dst_is_rgb24_or_bgr24) {
+        out_tx_csc->tx_csc_option = DMA2D_CSC_TX_SCRAMBLE; // RGB<->BGR conversion is just a scramble operation
+        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE0_1_2;
         return true;
     }
 
     if (src_fourcc == ESP_COLOR_FOURCC_RGB16 && dst_fourcc == ESP_COLOR_FOURCC_BGR24) {
         out_tx_csc->tx_csc_option = DMA2D_CSC_TX_RGB565_TO_RGB888;
-        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
-        out_tx_csc->post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
         return true;
     }
 
-    if (src_fourcc == ESP_COLOR_FOURCC_BGR24 && dst_fourcc == ESP_COLOR_FOURCC_RGB16) {
+    if (src_is_rgb24_or_bgr24 && dst_fourcc == ESP_COLOR_FOURCC_RGB16) {
         out_tx_csc->tx_csc_option = DMA2D_CSC_TX_RGB888_TO_RGB565;
-        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
-        out_tx_csc->post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
+        if (src_fourcc == ESP_COLOR_FOURCC_RGB24) {
+            out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE0_1_2;
+        }
         return true;
     }
 
-    if (src_fourcc == ESP_COLOR_FOURCC_BGR24 && dst_fourcc == ESP_COLOR_FOURCC_UYVY) {
+    if (src_is_rgb24_or_bgr24 && dst_fourcc == ESP_COLOR_FOURCC_UYVY) {
         if (request->color_conv_std == COLOR_CONV_STD_RGB_YUV_BT601) {
             out_tx_csc->tx_csc_option = DMA2D_CSC_TX_RGB888_TO_YUV422_601;
         } else if (request->color_conv_std == COLOR_CONV_STD_RGB_YUV_BT709) {
@@ -97,8 +129,9 @@ static bool try_convert_request_to_dma2d_csc(const async_color_convert_request_t
         } else {
             return false;
         }
-        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
-        out_tx_csc->post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
+        if (src_fourcc == ESP_COLOR_FOURCC_RGB24) {
+            out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE0_1_2;
+        }
         return true;
     }
 
@@ -110,8 +143,6 @@ static bool try_convert_request_to_dma2d_csc(const async_color_convert_request_t
         } else {
             return false;
         }
-        out_tx_csc->pre_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
-        out_tx_csc->post_scramble = DMA2D_SCRAMBLE_ORDER_BYTE2_1_0;
         return true;
     }
 
@@ -123,6 +154,11 @@ static inline bool needs_tx_csc(const dma2d_csc_config_t *tx_csc)
     return tx_csc->tx_csc_option != DMA2D_CSC_TX_NONE;
 }
 
+static inline bool needs_rx_csc(const dma2d_csc_config_t *rx_csc)
+{
+    return rx_csc->rx_csc_option != DMA2D_CSC_RX_NONE;
+}
+
 static esp_err_t sync_if_cacheable(void *addr, size_t size, int flags)
 {
     return esp_cache_get_line_size_by_addr(addr) > 0 ? esp_cache_msync(addr, size, flags) : ESP_OK;
@@ -239,9 +275,8 @@ static bool async_color_convert_on_job_picked(uint32_t channel_num,
     dma2d_set_transfer_ability(tx_chan, &transfer_ability);
     dma2d_set_transfer_ability(rx_chan, &transfer_ability);
 
-    if (needs_tx_csc(&trans->tx_csc)) {
-        dma2d_configure_color_space_conversion(tx_chan, &trans->tx_csc);
-    }
+    dma2d_configure_color_space_conversion(tx_chan, &trans->tx_csc);
+    dma2d_configure_color_space_conversion(rx_chan, &trans->rx_csc);
 
     dma2d_rx_event_callbacks_t cbs = {
         .on_recv_eof = async_color_convert_done_cb,
@@ -301,16 +336,18 @@ static esp_err_t async_color_convert_dma2d_convert(async_color_convert_context_t
 
     esp_color_fourcc_t src_fourcc = request->src_color_format;
     esp_color_fourcc_t dst_fourcc = request->dst_color_format;
-    trans->tx_csc = (dma2d_csc_config_t) {};
-    ESP_GOTO_ON_FALSE(try_convert_request_to_dma2d_csc(request, &trans->tx_csc),
+    trans->tx_csc = default_tx_csc_config();
+    trans->rx_csc = default_rx_csc_config();
+    ESP_GOTO_ON_FALSE(resolve_dma2d_csc_configs(request, &trans->tx_csc, &trans->rx_csc),
                       ESP_ERR_INVALID_ARG, recycle_and_out, TAG, "unsupported color conversion mode");
 
+    trans->dma2d_trans_config.channel_flags = DMA2D_CHANNEL_FUNCTION_FLAG_SIBLING;
     if (needs_tx_csc(&trans->tx_csc)) {
-        trans->dma2d_trans_config.channel_flags = DMA2D_CHANNEL_FUNCTION_FLAG_SIBLING | DMA2D_CHANNEL_FUNCTION_FLAG_TX_CSC;
-    } else {
-        trans->dma2d_trans_config.channel_flags = DMA2D_CHANNEL_FUNCTION_FLAG_SIBLING;
+        trans->dma2d_trans_config.channel_flags |= DMA2D_CHANNEL_FUNCTION_FLAG_TX_CSC;
+    }
+    if (needs_rx_csc(&trans->rx_csc)) {
+        trans->dma2d_trans_config.channel_flags |= DMA2D_CHANNEL_FUNCTION_FLAG_RX_CSC;
     }
-
     setup_desc(trans->tx_desc,
                (void *)request->src_buffer,
                request->src_stride,
@@ -342,7 +379,7 @@ static esp_err_t async_color_convert_dma2d_convert(async_color_convert_context_t
                       recycle_and_out, TAG, "source cache sync failed");
 
     ESP_GOTO_ON_ERROR(sync_if_cacheable(request->dst_buffer, dst_total_size,
-                                        ESP_CACHE_MSYNC_FLAG_DIR_C2M | ESP_CACHE_MSYNC_FLAG_INVALIDATE),
+                                        ESP_CACHE_MSYNC_FLAG_DIR_C2M | ESP_CACHE_MSYNC_FLAG_INVALIDATE | ESP_CACHE_MSYNC_FLAG_UNALIGNED),
                       recycle_and_out, TAG, "destination cache sync failed");
 
     ESP_GOTO_ON_ERROR(sync_if_cacheable(trans->tx_desc, color_ctx->desc_alloc_size,
diff --git a/components/esp_driver_dma/test_apps/dma2d/main/test_async_color_convert.c b/components/esp_driver_dma/test_apps/dma2d/main/test_async_color_convert.c
index ee2c500aa6f..01dd14ae605 100644
--- a/components/esp_driver_dma/test_apps/dma2d/main/test_async_color_convert.c
+++ b/components/esp_driver_dma/test_apps/dma2d/main/test_async_color_convert.c
@@ -48,10 +48,10 @@ TEST_CASE("async color convert basic callback", "[async_color_convert]")
 
     uint16_t *src565 = heap_caps_aligned_calloc(64, pixel_num, sizeof(uint16_t),
                                                 MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
-    uint8_t *dst888 = heap_caps_aligned_calloc(64, pixel_num, 3,
-                                               MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *dst_bgr24 = heap_caps_aligned_calloc(64, pixel_num, 3,
+                                                  MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
     TEST_ASSERT_NOT_NULL(src565);
-    TEST_ASSERT_NOT_NULL(dst888);
+    TEST_ASSERT_NOT_NULL(dst_bgr24);
 
     for (uint32_t i = 0; i < pixel_num; i++) {
         src565[i] = (uint16_t)((i * 13) ^ 0x5AA5);
@@ -71,7 +71,7 @@ TEST_CASE("async color convert basic callback", "[async_color_convert]")
         .src_height = height,
         .src_x = 0,
         .src_y = 0,
-        .dst_buffer = dst888,
+        .dst_buffer = dst_bgr24,
         .dst_stride = width,
         .dst_height = height,
         .dst_x = 0,
@@ -96,10 +96,10 @@ TEST_CASE("async color convert basic callback", "[async_color_convert]")
     TEST_ESP_OK(esp_async_color_convert_uninstall(conv_hdl));
 
     free(src565);
-    free(dst888);
+    free(dst_bgr24);
 }
 
-TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_convert]")
+TEST_CASE("async color convert roundtrip: RGB16<->BGR24", "[async_color_convert]")
 {
     const uint32_t width = 32;
     const uint32_t height = 20;
@@ -107,12 +107,12 @@ TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_conver
 
     uint16_t *src565 = heap_caps_aligned_calloc(64, pixel_num, sizeof(uint16_t),
                                                 MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
-    uint8_t *mid888 = heap_caps_aligned_calloc(64, pixel_num, 3,
-                                               MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *mid_bgr24 = heap_caps_aligned_calloc(64, pixel_num, 3,
+                                                  MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
     uint16_t *dst565 = heap_caps_aligned_calloc(64, pixel_num, sizeof(uint16_t),
                                                 MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
     TEST_ASSERT_NOT_NULL(src565);
-    TEST_ASSERT_NOT_NULL(mid888);
+    TEST_ASSERT_NOT_NULL(mid_bgr24);
     TEST_ASSERT_NOT_NULL(dst565);
 
     for (uint32_t i = 0; i < pixel_num; i++) {
@@ -127,13 +127,13 @@ TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_conver
     async_color_convert_handle_t conv_hdl = NULL;
     TEST_ESP_OK(esp_async_color_convert_install_dma2d(&config, &conv_hdl));
 
-    async_color_convert_request_t req_565_to_888 = {
+    async_color_convert_request_t req_565_to_bgr24 = {
         .src_buffer = src565,
         .src_stride = width,
         .src_height = height,
         .src_x = 0,
         .src_y = 0,
-        .dst_buffer = mid888,
+        .dst_buffer = mid_bgr24,
         .dst_stride = width,
         .dst_height = height,
         .dst_x = 0,
@@ -144,8 +144,8 @@ TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_conver
         .dst_color_format = ESP_COLOR_FOURCC_BGR24,
     };
 
-    async_color_convert_request_t req_888_to_565 = {
-        .src_buffer = mid888,
+    async_color_convert_request_t req_bgr24_to_565 = {
+        .src_buffer = mid_bgr24,
         .src_stride = width,
         .src_height = height,
         .src_x = 0,
@@ -161,8 +161,8 @@ TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_conver
         .dst_color_format = ESP_COLOR_FOURCC_RGB16,
     };
 
-    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_565_to_888, -1));
-    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_888_to_565, -1));
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_565_to_bgr24, -1));
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_bgr24_to_565, -1));
 
     // The final dst565 should be the same as the original src565 after round-trip conversion
     TEST_ASSERT_EQUAL_MEMORY(src565, dst565, pixel_num * sizeof(uint16_t));
@@ -170,7 +170,7 @@ TEST_CASE("async color convert roundtrip: RGB565<->RGB888", "[async_color_conver
     TEST_ESP_OK(esp_async_color_convert_uninstall(conv_hdl));
 
     free(src565);
-    free(mid888);
+    free(mid_bgr24);
     free(dst565);
 }
 
@@ -305,7 +305,189 @@ static void uyvy_to_bgr24_reference_image(const uint8_t *src_uyvy, uint8_t *dst_
     }
 }
 
-TEST_CASE("async color convert UYVY->RGB888 matches reference", "[async_color_convert]")
+TEST_CASE("async color convert swaps RGB24 and BGR24 byte order", "[async_color_convert]")
+{
+    const uint32_t width = 4;
+    const uint32_t height = 2;
+    const size_t pixel_count = width * height;
+    const size_t buf_size = pixel_count * 3;
+    static const uint8_t src_rgb24[] = {
+        0x10, 0x20, 0x30,  0x7F, 0x80, 0x81,  0xAA, 0x55, 0xFE,  0x01, 0xC0, 0x99,
+        0xDE, 0xAD, 0xBE,  0x00, 0x11, 0x22,  0x44, 0x88, 0xCC,  0xF0, 0x0D, 0x42,
+    };
+    static const uint8_t src_bgr24[] = {
+        0x30, 0x20, 0x10,  0x81, 0x80, 0x7F,  0xFE, 0x55, 0xAA,  0x99, 0xC0, 0x01,
+        0xBE, 0xAD, 0xDE,  0x22, 0x11, 0x00,  0xCC, 0x88, 0x44,  0x42, 0x0D, 0xF0,
+    };
+
+    TEST_ASSERT_EQUAL(sizeof(src_rgb24), buf_size);
+    TEST_ASSERT_EQUAL(sizeof(src_bgr24), buf_size);
+
+    uint8_t *rgb24 = heap_caps_aligned_calloc(64, 1, buf_size,
+                                              MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *bgr24 = heap_caps_aligned_calloc(64, 1, buf_size,
+                                              MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *dst_bgr24 = heap_caps_aligned_calloc(64, 1, buf_size,
+                                                  MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *dst_rgb24 = heap_caps_aligned_calloc(64, 1, buf_size,
+                                                  MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    TEST_ASSERT_NOT_NULL(rgb24);
+    TEST_ASSERT_NOT_NULL(bgr24);
+    TEST_ASSERT_NOT_NULL(dst_bgr24);
+    TEST_ASSERT_NOT_NULL(dst_rgb24);
+    memcpy(rgb24, src_rgb24, buf_size);
+    memcpy(bgr24, src_bgr24, buf_size);
+    memset(dst_bgr24, 0xA5, buf_size);
+    memset(dst_rgb24, 0x5A, buf_size);
+
+    async_color_convert_config_t config = {
+        .backlog = 1,
+        .intr_priority = 0,
+        .dma_burst_size = 16,
+    };
+    async_color_convert_handle_t conv_hdl = NULL;
+    TEST_ESP_OK(esp_async_color_convert_install_dma2d(&config, &conv_hdl));
+
+    async_color_convert_request_t req_rgb_to_bgr = {
+        .src_buffer = rgb24,
+        .src_stride = width,
+        .src_height = height,
+        .src_x = 0,
+        .src_y = 0,
+        .dst_buffer = dst_bgr24,
+        .dst_stride = width,
+        .dst_height = height,
+        .dst_x = 0,
+        .dst_y = 0,
+        .copy_width = width,
+        .copy_height = height,
+        .src_color_format = ESP_COLOR_FOURCC_RGB24,
+        .dst_color_format = ESP_COLOR_FOURCC_BGR24,
+    };
+    async_color_convert_request_t req_bgr_to_rgb = {
+        .src_buffer = bgr24,
+        .src_stride = width,
+        .src_height = height,
+        .src_x = 0,
+        .src_y = 0,
+        .dst_buffer = dst_rgb24,
+        .dst_stride = width,
+        .dst_height = height,
+        .dst_x = 0,
+        .dst_y = 0,
+        .copy_width = width,
+        .copy_height = height,
+        .src_color_format = ESP_COLOR_FOURCC_BGR24,
+        .dst_color_format = ESP_COLOR_FOURCC_RGB24,
+    };
+
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_rgb_to_bgr, -1));
+    TEST_ASSERT_EQUAL_MEMORY(src_bgr24, dst_bgr24, buf_size);
+
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_bgr_to_rgb, -1));
+    TEST_ASSERT_EQUAL_MEMORY(src_rgb24, dst_rgb24, buf_size);
+
+    TEST_ESP_OK(esp_async_color_convert_uninstall(conv_hdl));
+    free(rgb24);
+    free(bgr24);
+    free(dst_bgr24);
+    free(dst_rgb24);
+}
+
+// Verifies the scramble route and BGR24/RGB24->UYVY conversion compose correctly.
+TEST_CASE("async color convert RGB24 and BGR24 inputs produce identical UYVY output", "[async_color_convert]")
+{
+    const uint32_t width = 4;
+    const uint32_t height = 2;
+    const size_t pixel_count = width * height;
+    const size_t rgb_size = pixel_count * 3;
+    const size_t uyvy_size = pixel_count * 2;
+    static const uint8_t src_rgb24[] = {
+        0x10, 0x20, 0x30,  0x7F, 0x80, 0x81,  0xAA, 0x55, 0xFE,  0x01, 0xC0, 0x99,
+        0xDE, 0xAD, 0xBE,  0x00, 0x11, 0x22,  0x44, 0x88, 0xCC,  0xF0, 0x0D, 0x42,
+    };
+    static const uint8_t src_bgr24[] = {
+        0x30, 0x20, 0x10,  0x81, 0x80, 0x7F,  0xFE, 0x55, 0xAA,  0x99, 0xC0, 0x01,
+        0xBE, 0xAD, 0xDE,  0x22, 0x11, 0x00,  0xCC, 0x88, 0x44,  0x42, 0x0D, 0xF0,
+    };
+    const color_conv_std_rgb_yuv_t conv_std = COLOR_CONV_STD_RGB_YUV_BT601;
+
+    TEST_ASSERT_EQUAL(sizeof(src_rgb24), rgb_size);
+    TEST_ASSERT_EQUAL(sizeof(src_bgr24), rgb_size);
+
+    uint8_t *rgb24 = heap_caps_aligned_calloc(64, 1, rgb_size,
+                                              MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *bgr24 = heap_caps_aligned_calloc(64, 1, rgb_size,
+                                              MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *dst_from_rgb24 = heap_caps_aligned_calloc(64, 1, uyvy_size,
+                                                       MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    uint8_t *dst_from_bgr24 = heap_caps_aligned_calloc(64, 1, uyvy_size,
+                                                       MALLOC_CAP_INTERNAL | MALLOC_CAP_DMA | MALLOC_CAP_8BIT);
+    TEST_ASSERT_NOT_NULL(rgb24);
+    TEST_ASSERT_NOT_NULL(bgr24);
+    TEST_ASSERT_NOT_NULL(dst_from_rgb24);
+    TEST_ASSERT_NOT_NULL(dst_from_bgr24);
+    memcpy(rgb24, src_rgb24, rgb_size);
+    memcpy(bgr24, src_bgr24, rgb_size);
+
+    async_color_convert_config_t config = {
+        .backlog = 2,
+        .intr_priority = 0,
+        .dma_burst_size = 16,
+    };
+    async_color_convert_handle_t conv_hdl = NULL;
+    TEST_ESP_OK(esp_async_color_convert_install_dma2d(&config, &conv_hdl));
+
+    memset(dst_from_rgb24, 0xA5, uyvy_size);
+    memset(dst_from_bgr24, 0x5A, uyvy_size);
+
+    async_color_convert_request_t req_rgb24_to_uyvy = {
+        .src_buffer = rgb24,
+        .src_stride = width,
+        .src_height = height,
+        .src_x = 0,
+        .src_y = 0,
+        .dst_buffer = dst_from_rgb24,
+        .dst_stride = width,
+        .dst_height = height,
+        .dst_x = 0,
+        .dst_y = 0,
+        .copy_width = width,
+        .copy_height = height,
+        .src_color_format = ESP_COLOR_FOURCC_RGB24,
+        .dst_color_format = ESP_COLOR_FOURCC_UYVY,
+        .color_conv_std = conv_std,
+    };
+    async_color_convert_request_t req_bgr24_to_uyvy = {
+        .src_buffer = bgr24,
+        .src_stride = width,
+        .src_height = height,
+        .src_x = 0,
+        .src_y = 0,
+        .dst_buffer = dst_from_bgr24,
+        .dst_stride = width,
+        .dst_height = height,
+        .dst_x = 0,
+        .dst_y = 0,
+        .copy_width = width,
+        .copy_height = height,
+        .src_color_format = ESP_COLOR_FOURCC_BGR24,
+        .dst_color_format = ESP_COLOR_FOURCC_UYVY,
+        .color_conv_std = conv_std,
+    };
+
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_rgb24_to_uyvy, -1));
+    TEST_ESP_OK(esp_color_convert_blocking(conv_hdl, &req_bgr24_to_uyvy, -1));
+    TEST_ASSERT_EQUAL_MEMORY(dst_from_bgr24, dst_from_rgb24, uyvy_size);
+
+    TEST_ESP_OK(esp_async_color_convert_uninstall(conv_hdl));
+    free(rgb24);
+    free(bgr24);
+    free(dst_from_rgb24);
+    free(dst_from_bgr24);
+}
+
+TEST_CASE("async color convert UYVY->BGR24 matches reference", "[async_color_convert]")
 {
     const uint32_t src_stride = 32;
     const uint32_t dst_stride = 64;
diff --git a/components/esp_hal_dma/esp32p4/include/hal/dma2d_ll.h b/components/esp_hal_dma/esp32p4/include/hal/dma2d_ll.h
index f96263f8e61..4eb70c39136 100644
--- a/components/esp_hal_dma/esp32p4/include/hal/dma2d_ll.h
+++ b/components/esp_hal_dma/esp32p4/include/hal/dma2d_ll.h
@@ -1054,7 +1054,7 @@ static inline void dma2d_ll_tx_configure_color_space_conv(dma2d_dev_t *dev, uint
         input_sel = 7;
         break;
     case DMA2D_CSC_TX_SCRAMBLE:
-        input_sel = 2; // Or 3
+        input_sel = 3; // Other 3-byte/pixel input path
         proc_en = false;
         output_sel = 2;
         break;
diff --git a/components/esp_hal_dma/esp32s31/include/hal/dma2d_ll.h b/components/esp_hal_dma/esp32s31/include/hal/dma2d_ll.h
index 27970ecf257..9e8921d447f 100644
--- a/components/esp_hal_dma/esp32s31/include/hal/dma2d_ll.h
+++ b/components/esp_hal_dma/esp32s31/include/hal/dma2d_ll.h
@@ -1031,7 +1031,7 @@ static inline void dma2d_ll_tx_configure_color_space_conv(dma2d_dev_t *dev, uint
         input_sel = 7;
         break;
     case DMA2D_CSC_TX_SCRAMBLE:
-        input_sel = 2; // Or 3
+        input_sel = 3; // Other 3-byte/pixel input path
         proc_en = false;
         output_sel = 2;
         break;
diff --git a/components/esp_lcd/dsi/esp_lcd_panel_dpi.c b/components/esp_lcd/dsi/esp_lcd_panel_dpi.c
index 3c38f45aca2..14ec95ec3b3 100644
--- a/components/esp_lcd/dsi/esp_lcd_panel_dpi.c
+++ b/components/esp_lcd/dsi/esp_lcd_panel_dpi.c
@@ -533,7 +533,9 @@ esp_err_t esp_lcd_dpi_panel_enable_dma2d(esp_lcd_panel_handle_t panel)
 
     // Initialize the async color convert backend used by the built-in DMA2D copy hook.
     // Use its default backlog to queue multiple frame buffer copy requests.
-    async_color_convert_config_t fbcpy_config = {};
+    async_color_convert_config_t fbcpy_config = {
+        .dma_burst_size = 128, // for better performance
+    };
     ESP_RETURN_ON_ERROR(esp_async_color_convert_install_dma2d(&fbcpy_config, &dpi_panel->fbcpy_handle), TAG, "install async frame buffer copy backend failed");
 
     // Register the DMA2D draw bitmap hook
diff --git a/docs/en/api-reference/peripherals/async_color_convert.rst b/docs/en/api-reference/peripherals/async_color_convert.rst
index 548468053ab..7980144afa6 100644
--- a/docs/en/api-reference/peripherals/async_color_convert.rst
+++ b/docs/en/api-reference/peripherals/async_color_convert.rst
@@ -148,7 +148,7 @@ Both the source window and destination window must stay within the bounds of the
 Supported Conversions
 ---------------------
 
-The following format pairs are supported by this driver:
+The following format pairs are currently supported by this driver:
 
 .. list-table::
     :header-rows: 1
@@ -156,32 +156,54 @@ The following format pairs are supported by this driver:
     * - Source format
       - Destination format
       - Conversion standard
-    * - same as destination (skip conversion)
-      - same as source (skip conversion)
+    * - ``ESP_COLOR_FOURCC_RGB16``
+      - ``ESP_COLOR_FOURCC_RGB16``
       - N/A
-    * - RGB565
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_BGR24``
       - N/A
-    * - RGB888
-      - RGB565
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_RGB24``
       - N/A
-    * - RGB888
-      - UYVY422
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_UYVY``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_RGB24``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_RGB16``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_RGB16``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_RGB16``
+      - N/A
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.601
-    * - RGB888
-      - UYVY422
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.709
-    * - UYVY422
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.601
-    * - UYVY422
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_UYVY``
+      - BT.709
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - BT.601
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_BGR24``
       - BT.709
 
 .. note::
 
-    In this driver, RGB888 uses ``ESP_COLOR_FOURCC_BGR24`` and UYVY422 uses ``ESP_COLOR_FOURCC_UYVY``.
-
     Always set :cpp:member:`async_color_convert_request_t::src_color_format` and
     :cpp:member:`async_color_convert_request_t::dst_color_format`.
     Set :cpp:member:`async_color_convert_request_t::color_conv_std` when converting between RGB and YUV.
diff --git a/docs/zh_CN/api-reference/peripherals/async_color_convert.rst b/docs/zh_CN/api-reference/peripherals/async_color_convert.rst
index 21041f6b634..32a62dd9aee 100644
--- a/docs/zh_CN/api-reference/peripherals/async_color_convert.rst
+++ b/docs/zh_CN/api-reference/peripherals/async_color_convert.rst
@@ -148,7 +148,7 @@
 支持的转换格式
 --------------
 
-本驱动支持以下格式组合：
+本驱动当前支持以下格式组合：
 
 .. list-table::
     :header-rows: 1
@@ -156,32 +156,54 @@
     * - 源格式
       - 目标格式
       - 转换标准
-    * - 与目标格式相同（跳过转换）
-      - 与源格式相同（跳过转换）
+    * - ``ESP_COLOR_FOURCC_RGB16``
+      - ``ESP_COLOR_FOURCC_RGB16``
       - 不适用
-    * - RGB565
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_BGR24``
       - 不适用
-    * - RGB888
-      - RGB565
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_RGB24``
       - 不适用
-    * - RGB888
-      - UYVY422
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_UYVY``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_RGB24``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_RGB16``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_RGB16``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_RGB16``
+      - 不适用
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.601
-    * - RGB888
-      - UYVY422
+    * - ``ESP_COLOR_FOURCC_BGR24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.709
-    * - UYVY422
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_UYVY``
       - BT.601
-    * - UYVY422
-      - RGB888
+    * - ``ESP_COLOR_FOURCC_RGB24``
+      - ``ESP_COLOR_FOURCC_UYVY``
+      - BT.709
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_BGR24``
+      - BT.601
+    * - ``ESP_COLOR_FOURCC_UYVY``
+      - ``ESP_COLOR_FOURCC_BGR24``
       - BT.709
 
 .. note::
 
-    在本驱动中，RGB888 使用 ``ESP_COLOR_FOURCC_BGR24``，UYVY422 使用 ``ESP_COLOR_FOURCC_UYVY``。
-
     所有请求都需要设置 :cpp:member:`async_color_convert_request_t::src_color_format` 和
     :cpp:member:`async_color_convert_request_t::dst_color_format`。
     当在 RGB 和 YUV 之间转换时，还需要设置 :cpp:member:`async_color_convert_request_t::color_conv_std`。