Issue with dm365 and ov9653

Hi Lincoln:

            Have you captured raw data and have a view?

            We need focus where the problem would happened firstly, Capture? Encode? or Decoder?

            If you can find the same issue on raw data, better to attachmed a raw data for reference.

            I guess something wrong with the sensor.

Hi,

What are the OV exposure and gain set to? Are you using AEC and AGC?

You may be looking at artifacts produced by high sensor gain.

Try taking a video using only ambient light from a brighter source (preferably incandescent light bulb.) Do not use the LEDs, use colored paper or a colorful magazine cover or something else instead. Or better yet, take a video of a color checker chart outside on a cloudy day.

JPM

Hi Tracy:

It's captured raw data, hasn't compressed yet.

Hi JPM:

Thanks for your quick reply.

I have done some experiments about AEC and AGC on ov9653, unfortunately it takes no effect for this issue.

I also used different enviroments to test, the strips exists always.

Additionally, I fixed both U and V values to 0x80 , and captured again. I got the expected raw data with U and V set to 0x80. 

So can I say that , settings over DM365 is correct, this issue is caused by wrong settings over ov9653?

Following is the raw data I captured with U and V set to 0x80:

2234.UYVY_8BIT_640x480_black.zip

We have solved this problem.

It's cause by the unstable power supply.

After add a Tantalum capacitors to AVDD, the stripes disappeared.

Hi Loncoln,

I'm glad you found your problem! I'm curious, are you using a switching supply or LDO for the AVDD? I am also curious what value/number of capacitors were used for AVDD before and after the fix?

Regards,

JPM

Hi JPM

We are using LDO for AVDD.

Replaced the 10uF capacitor to tantalum type.