build a captured picture file example

[jasperchih]

Hi,

i completed to build the ESP32-CAM support micropython bin file and loaded to the board. The current consumption looks reasonable. But, no idea on how to capture the image and save into a file. Can you help on an example script ? Or, please advise how to check out the supported class function. thanks million.

jasper

[tsaarni]

Hi,

There is only single function in the camera module called capture() which captures JPG image and returns that as a buffer. You can see it being used in the web server script webcam.py. It returns the image as HTTP response, but it could just as well be saved into a file.

The capture() function currently does not take any parameters, so there is no control over the detailed sensor settings. You can see the C implementation calling the camera driver here with those hardcoded sensor settings.

[rychong]

Hi, Can you send me a copy of your bin file? the email is renyichong@outlook.com

[tsaarni]

I uploaded firmware image here https://github.com/tsaarni/esp32-micropython-webcam/releases/tag/2019-02-24

[jasperchih]

Hi, thanks for the upload... i measure the power consumption result (shown in below) and looks very close with my compiled code. i feel exciting and appreciate on your notes. but,, the remain question is : how to create the camera settings Python method? For example, "camera.resolution(XXXX)". Please let me know which file to start with if possible.

thanks

[tsaarni]

The methods can be implemented in this file https://github.com/tsaarni/micropython-with-esp32-cam/blob/esp32-camera-for-micropython/ports/esp32/modcamera.c by following the example of existing methods. I used this document as reference how to add methods to micropython. You'll find also how to add methods with arguments, that is, how to convert between Python and C data types. Rest of the type conversion methods are defined here.

The functions to control the sensor seem to be listed here. To controlling resolution, I only find set_framesize() which takes enumeration framesize_t argument (defined in the same file). I'm not sure though, if this is the only way to control resolution.

So I think minimum viable code could look something along the lines like this

STATIC mp_obj_t camera_set_framesize(mp_obj_t framesize_obj){
    framesize_t framesize_enum = mp_obj_int_get_truncated(framesize_obj);
    sensor_t *control = esp_camera_sensor_get();
    control->set_framesize(control, framesize_enum);  
    return mp_const_true;
}
STATIC MP_DEFINE_CONST_FUN_OBJ_1(camera_set_framesize_obj, camera_set_framesize);

STATIC const mp_rom_map_elem_t camera_module_globals_table[] = {
    // all the existing methods remain also here but add following....
    { MP_ROM_QSTR(MP_QSTR_capture), MP_ROM_PTR(&camera_set_framesize_obj) },
}

Of course this would leave lot to be desired from usability and error handling. Note: I have not tested this, so it probably does not even compile, but I hope it provides a starting point anyway!

[tsaarni]

I checked the resolution issue a bit more and I suspect that the sensor has fixed resolution options. So maybe rest would need to be done in software.

If you look at the driver the set_framesize() function for ov2640 supports only three options out of the framesize_t enum: UXGA, SVGA, CIF. The information in the datasheet of ov2640 seems to match (see bits for resolution selection in table 13 on page 24).