Issue with capture_continuous

[weathernick]

I think I found a bug or a feature?

I run capture_continuous and have it pause for 6ish seconds. I just noticed when I upgraded to 1.12 that back to back images are the same. So for time 0sec I have an images, 6 seconds later I grab another image from capture_continuous and it turns out to be the same image. 6 seconds later I get a new image but it gets doubled on the next cycle. so instead of ABCDEFG I get AACCEEGG.

My code:
for i, filename in enumerate(camera.capture_continuous(nightpath+'/{timestamp:%H%M%S}.jpg', resize=resolution, burst=True, quality=quality)): sleep = nextpictime("night",filename) time.sleep(sleep)

I tried turning off burst and addjusting my sleep but same results. I reverted back to 1.10 which works fine for me and my four weather cameras.

Just figured I would let you know.

Thanks,

Nickolas

[waveform80]

What an intriguing bug! Confirmed on 1.12 with the following replication code:

import os
import picamera
import time

nightpath = '.'
camera = picamera.PiCamera(resolution=(1024, 768))
camera.start_preview()

for i, filename in enumerate(
        camera.capture_continuous(
            os.path.join(nightpath, '{timestamp:%H%M%S}.jpg'),
            resize=(640, 480), burst=True, quality=90)):
        sleep = 1
        time.sleep(sleep)

And some captures to demonstrate the duplication (not sure what's going on with 094800.jpg - something weird). I'll try and have a look at this next week.

[waveform80]

Hmm, that took some work but I've got this fixed up now. Many thanks for mentioning which version worked - that helped a great deal (although unfortunately a huge amount changed under the covers from 1.10 to 1.11 which took some time to dig through - anyway, it was indeed one of those changes that caused the issue).

[alexellis]

I feel like I'm seeing the same issue with this code:

  engine = DetectionEngine(args.model)
  labels = ReadLabelFile(args.label) if args.label else None

  with picamera.PiCamera() as camera:
    camera.resolution = (640, 480)
    camera.framerate = 30
    _, width, height, channels = engine.get_input_tensor_shape()
    camera.start_preview()
    try:
        stream = io.BytesIO()
        for foo in camera.capture_continuous(stream,
                                                format='jpeg',
                                                use_video_port=True,
                                                resize=(width, height)):
            stream.truncate()
            stream.seek(0)
            img = Image.open(stream)

            ans = engine.DetectWithImage(img, threshold=0.1, keep_aspect_ratio=True,
                                            relative_coord=False, top_k=10)
            img=None
            print("ans", ans)     
            if ans:
                for obj in ans:
                    print ('-----------------------------------------')
                    if labels:
                        print(labels[obj.label_id])
                        print ('score = ', obj.score)
    finally:
        camera.stop_preview()
        stream.close()

I can only get different classifications if I close the program and re-open it.

What would you advise? I appear to have "picamera 1.13"

Alex