diff --git a/ili9341.c b/ili9341.c
index cb5e0ec..6fbb2c4 100644
--- a/ili9341.c
+++ b/ili9341.c
@@ -52,6 +52,18 @@ ssp_senddata(uint8_t x)
     ;
 }
 
+uint8_t
+ssp_sendrecvdata(uint8_t x)
+{
+    while (!(SPI1->SR & SPI_SR_TXE));
+    // clear OVR
+    while (SPI1->SR & SPI_SR_RXNE) (void)SPI1->DR;
+
+    *(uint8_t*)(&SPI1->DR) = x;
+    while (!(SPI1->SR & SPI_SR_RXNE));
+    return SPI1->DR;
+}
+
 void
 ssp_senddata16(uint16_t x)
 {
@@ -104,8 +116,8 @@ spi_init(void)
   dmaStreamSetPeripheral(dmatx, &SPI1->DR);
 
   SPI1->CR1 = 0;
-  SPI1->CR1 = SPI_CR1_MSTR | SPI_CR1_BIDIOE | SPI_CR1_SSM | SPI_CR1_SSI;// | SPI_CR1_BR_1;
-  SPI1->CR2 = 0x0700 | SPI_CR2_TXDMAEN;
+  SPI1->CR1 = SPI_CR1_MSTR | SPI_CR1_SSM | SPI_CR1_SSI;// | SPI_CR1_BR_1;
+  SPI1->CR2 = 0x0700 | SPI_CR2_TXDMAEN | SPI_CR2_FRXTH;
   SPI1->CR1 |= SPI_CR1_SPE;  
 }
 
@@ -278,6 +290,50 @@ void ili9341_bulk(int x, int y, int w, int h)
 }
 #endif
 
+void
+ili9341_read_memory_raw(uint8_t cmd, int len, uint16_t* out)
+{
+    uint8_t r, g, b;
+    send_command(cmd, 0, NULL);
+    ssp_databit8();
+
+    // consume old data
+    while (!(SPI1->SR & SPI_SR_TXE));
+    // clear OVR
+    while (SPI1->SR & SPI_SR_RXNE) r = SPI1->DR;
+
+    // require 8bit dummy clock
+    r = ssp_sendrecvdata(0);
+
+    while (len-- > 0) {
+        // read data is always 18bit
+        r = ssp_sendrecvdata(0);
+        g = ssp_sendrecvdata(0);
+        b = ssp_sendrecvdata(0);
+        *out++ = ((r & 0xF8) << 8) | ((g & 0xFC) << 3) | (b >> 3);
+    }
+
+    CS_HIGH;
+}
+
+void
+ili9341_read_memory(int x, int y, int w, int h, int len, uint16_t *out)
+{
+    uint8_t xx[4] = { x >> 8, x, (x+w-1) >> 8, (x+w-1) };
+    uint8_t yy[4] = { y >> 8, y, (y+h-1) >> 8, (y+h-1) };
+
+    send_command(0x2A, 4, xx);
+    send_command(0x2B, 4, yy);
+
+    ili9341_read_memory_raw(0x2E, len, out);
+}
+
+void
+ili9341_read_memory_continue(int len, uint16_t* out)
+{
+    ili9341_read_memory_raw(0x3E, len, out);
+}
+
 void
 ili9341_drawchar_5x7(uint8_t ch, int x, int y, uint16_t fg, uint16_t bg)
 {
diff --git a/main.c b/main.c
index ae68e19..f18d7dc 100644
--- a/main.c
+++ b/main.c
@@ -51,6 +51,7 @@ int8_t frequency_updated = FALSE;
 int8_t sweep_enabled = TRUE;
 int8_t cal_auto_interpolate = TRUE;
 int8_t redraw_requested = FALSE;
+int8_t stop_the_world = FALSE;
 
 static THD_WORKING_AREA(waThread1, 768);
 static THD_FUNCTION(Thread1, arg)
@@ -59,6 +60,11 @@ static THD_FUNCTION(Thread1, arg)
     chRegSetThreadName("sweep");
 
     while (1) {
+      if (stop_the_world) {
+          __WFI();
+          continue;
+      }
+
       if (sweep_enabled) {
         chMtxLock(&mutex);
         sweep();
@@ -377,6 +383,45 @@ static void cmd_dump(BaseSequentialStream *chp, int argc, char *argv[])
 }
 #endif
 
+static void cmd_capture(BaseSequentialStream *chp, int argc, char *argv[])
+{
+// read pixel count at one time (PART*2 bytes required for read buffer)
+#define PART 320
+    (void)argc;
+    (void)argv;
+
+    chMtxLock(&mutex);
+
+    // pause sweep
+    stop_the_world = TRUE;
+
+    chThdSleepMilliseconds(1000);
+
+    // use uint16_t spi_buffer[1024] (defined in ili9341) for read buffer
+    uint16_t *buf = &spi_buffer[0];
+    int len = 320 * 240;
+    int i;
+    ili9341_read_memory(0, 0, 320, 240, PART, buf);
+    for (i = 0; i < PART; i++) {
+        streamPut(chp, buf[i] >> 8);
+        streamPut(chp, buf[i] & 0xff);
+    }
+
+    len -= PART;
+    while (len > 0) {
+        ili9341_read_memory_continue(PART, buf);
+        for (i = 0; i < PART; i++) {
+            streamPut(chp, buf[i] >> 8);
+            streamPut(chp, buf[i] & 0xff);
+        }
+        len -= PART;
+    }
+    //*/
+
+    stop_the_world = FALSE;
+    chMtxUnlock(&mutex);
+}
+
 #if 0
 static void cmd_gamma(BaseSequentialStream *chp, int argc, char *argv[])
 {
@@ -1713,6 +1758,7 @@ static const ShellCommand commands[] =
     { "trace", cmd_trace },
     { "marker", cmd_marker },
     { "edelay", cmd_edelay },
+    { "capture", cmd_capture },
     { NULL, NULL }
 };
 
diff --git a/nanovna.h b/nanovna.h
index 1ef3ede..e1ea1c6 100644
--- a/nanovna.h
+++ b/nanovna.h
@@ -256,6 +256,8 @@ void ili9341_fill(int x, int y, int w, int h, int color);
 void ili9341_drawchar_5x7(uint8_t ch, int x, int y, uint16_t fg, uint16_t bg);
 void ili9341_drawstring_5x7(const char *str, int x, int y, uint16_t fg, uint16_t bg);
 void ili9341_drawfont(uint8_t ch, const font_t *font, int x, int y, uint16_t fg, uint16_t bg);
+void ili9341_read_memory(int x, int y, int w, int h, int len, uint16_t* out);
+void ili9341_read_memory_continue(int len, uint16_t* out);
 
 
 /*
diff --git a/python/nanovna.py b/python/nanovna.py
index 5fd9e89..aa70ef8 100755
--- a/python/nanovna.py
+++ b/python/nanovna.py
@@ -4,6 +4,7 @@ import numpy as np
 import pylab as pl
 import scipy.signal as signal
 import time
+import struct
 
 REF_LEVEL = (1<<9)
 
@@ -188,6 +189,16 @@ class NanoVNA():
                 x.append(float(line))
         self._frequencies = np.array(x)
 
+    def capture(self):
+        from PIL import Image
+        self.send_command("capture\r")
+        b = self.serial.read(320 * 240 * 2)
+        x = struct.unpack(">76800H", b)
+        # convert pixel format from 565(RGB) to 8888(RGBA)
+        arr = np.array(x, dtype=np.uint32)
+        arr = 0xFF000000 + ((arr & 0xF800) >> 8) + ((arr & 0x07E0) << 5) + ((arr & 0x001F) << 19)
+        return Image.frombuffer('RGBA', (320, 240), arr, 'raw', 'RGBA', 0, 1)
+
     def logmag(self, x):
         pl.grid(True)
         pl.plot(self.frequencies, 20*np.log10(np.abs(x)))
@@ -323,9 +334,18 @@ if __name__ == '__main__':
     parser.add_option("-l", "--filter",
                       action="store_true", dest="filter", default=False,
                       help="apply IF filter on raw wave plot")
+    parser.add_option("-C", "--capture", dest="capture",
+                      help="capture current display to FILE", metavar="FILE")
     (opt, args) = parser.parse_args()
 
     nv = NanoVNA(opt.device or '/dev/cu.usbmodem401')
+
+    if opt.capture:
+        print("capturing...")
+        img = nv.capture()
+        img.save(opt.capture)
+        exit(0)
+
     nv.set_frequency(opt.freq)
     nv.set_port(opt.port)
     nv.set_gain(opt.gain)