STICKY: Foundation (old) Camera Board information

[karne]

Hello,

Can anybody tell what is set in OV5647 MIPI registers (0x4800 to 0x4867) to make it happily communicate with RaspPI?
Possibly this can be just read out by hooking on I2C but camera modules are out of stock at the moment and this is not a GPU so, hopefully not a secret...
Knowing this we could make more compatible devices (aka cameras)

[jbeale]

Can anybody tell what is set in OV5647 MIPI registers (0x4800 to 0x4867) to make it happily communicate with RaspPI?
Possibly this can be just read out by hooking on I2C but camera modules are out of stock at the moment and this is not a GPU so, hopefully not a secret...
Knowing this we could make more compatible devices (aka cameras)

I have a camera and might be able to log the I2C traffic with a logic analyzer. But would this help you? Do you have the full datasheet for the OV5647 including the registers? Is that online somewhere? I kind of like the idea of a closed-source-GPU workaround by coming up with a new device that pretends to be an OV5647.

[jamesh]

Hello,

Can anybody tell what is set in OV5647 MIPI registers (0x4800 to 0x4867) to make it happily communicate with RaspPI?
Possibly this can be just read out by hooking on I2C but camera modules are out of stock at the moment and this is not a GPU so, hopefully not a secret...
Knowing this we could make more compatible devices (aka cameras)

The datasheet I have which has this stuff in is Confidential to Broadcom. I think you will need to approach Omnivision directly.

Note that we also set a whole load of other registers outside that range to make it all work.

[karne]

I have the full OV5647 datasheet. Also have full MIPI CSI-2 specs doc. Contact me with your email address I will send it.
Yes, if possible pls log I2C on MIPI and look how it communicates with camera.
This is what I was intended to do but cant get a camera at the moment.
I'm so surprised why the MIPI interface is so secret? This has nothing to do with Bradcom chip - just a standard interface(fast!!!) to outside world. It can be used for limitless imaging applications. Also I dont understand why not to expose the I2C out of the driver along with the MIPI configuration functions. The only explanation comes to head is to force community to buy more original RaspiCamers and not allow others.

[jamesh]

The CSI-2 connection goes straight into the GPU which is closed source. On there reside the camera driver, the camera tuning and the ISP, and all require access to the source and the GPU CPU compilers (its a custom instruction set) to make any changes. So you need to replicate exactly the behaviour of the OV5647, otherwise it's is very unlikely to work.

And if you replicate it exactly (i.e. same timings and resolution), you might as well use the 5647.

[karne]

Ok, James, it looks you are the one who knows all that. And - yes, exactly as you say we want to replicated OV5647 end. Not the Raspi GPU. Can you share with us info about what settings are used in OV5647 when it is talking to Raspi.
Like, bus speed, timings, both lanes used or one, video format and other settings so we could replicate its behavior.
This is what was initial question on the top this thread.

[jbeale]

So you need to replicate exactly the behaviour of the OV5647, otherwise it's is very unlikely to work. And if you replicate it exactly (i.e. same timings and resolution), you might as well use the 5647.

My understanding was the OP wanted to construct a different type of device, which is quite different from a mobile phone camera, but which happens to generate a bitmap stream and presents the OV5647-like control and data interface purely as a workaround to the closed-source nature of the GPU side. So it could work with the existing RPi board, firmware, and software as-is. I don't know enough to say whether this is possible or not. If this could be done, it seems like a quite excellent hack to me.

[jamesh]

Ok, James, it looks you are the one who knows all that. And - yes, exactly as you say we want to replicated OV5647 end. Not the Raspi GPU. Can you share with us info about what settings are used in OV5647 when it is talking to Raspi.
Like, bus speed, timings, both lanes used or one, video format and other settings so we could replicate its behavior.
This is what was initial question on the top this thread.

I do not know what I am allowed to publish and what I am not allowed to publish. I'll try and find out.

[peepo]

is there an update on available width*height*fps formats?
in particular:

any higher than 30fps

320*240 format,
ie not expecting faster than 30-90fps,
rather cpu cycles available

best

~:"

[jamesh]

is there an update on available width*height*fps formats?
in particular:

any higher than 30fps

320*240 format,
ie not expecting faster than 30-90fps,
rather cpu cycles available

best

~:"

No, just the stuff reported. I'll try and update here for any fixes released.

[jamesh]

OK, we released a few fixes recently (require new app code and in some cases firmware as well)

RaspiStill timelapse mode frames going Dark : Fixed.
RaspiStill timelapse mode not adding raw data to captures after the first : Fixed
RaspiStill EXIF data missing : Fixed
RaspiStill ISO setting not working : Fixed

All- Bad maze like artifacts in lens flare areas : Fixed (slight degradation in sharpness as a consequence - no way round this as is a sensor issue we need to get round).

Plus some new features :

RaspiStill timelapse mode missing : added
RaspiStillYUV : Now supports RGB888 output
RaspiStill : Added a preview mode that runs in the capture resolution ie the camera works at full res. Runs up to 15fps, but FOV of capture is same as preview. Hoping it's slightly faster for captures, but still needs testing.

I've not updated the readme's yet.

Have fun!


In other news, today I figured out why the preview disappears in RaspiStill when you set h or v flip (took a whole day!). Should be released in next firmware release - hopefully tonight or tomorrow.

James

[towolf]

Nice progress. Thank you James.

Any news on full-frame (binned) video?

[dom]

In other news, today I figured out why the preview disappears in RaspiStill when you set h or v flip (took a whole day!). Should be released in next firmware release - hopefully tonight or tomorrow.

The firmware is pushed. rpi-update to get it.

[jamesh]

Nice progress. Thank you James.

Any news on full-frame (binned) video?

Nope. Not started on that yet. Been spending too much time on Raspi stuff recently so will have to do some actual work for while.

I'll post here when I have any releasable changes.

[rodneycolton]

I used the "Unofficial guide to getting ...." the raspi cam up and running.
The guide refers to online documentation for more details about all of the commands and options.
I've searched, but can't find it.
Is there a complete Users Manual for the raspicam with more detail than the "unofficial guide"?

Thanks
Rodney Colton

[ripping]

Is there a complete Users Manual for the raspicam with more detail than the "unofficial guide"?
Thanks
Rodney Colton

I've been looking for the very same thing for the last hour or so.
I found this which expands on what is given from the command line with raspivid and raspistill.

https://github.com/raspberrypi/userland ... amDocs.odt

[poing]

This?
https://github.com/JamesH65/userland/tr ... iew-window

[jamesh]

The latest version of the Magpi has an article all about setting up the camera board. Next issue with continue with more detail.

The .odt file in the userland repo describes all the command line options (I updated it last week)

I think it need to be in a more prominent place actually (and could do with some more detail and better formatting), I'll chat with Liz about it.

[recantha2]

I'll have a go at reformatting if it'll help.

[jamesh]

I'm mid way through a reformat anyway, and I want to add some extra stuff on setup, but thanks for the offer.

[liz]

The user documentation is now available as a pdf - enjoy!

[wallarug]

The user documentation is now available as a pdf - enjoy!

Thanks liz!

[deSitter]

Thanks for all the work you have done jamesh. I wish I could loan you all of my free so you could implement the higher fps modes

[jamesh]

Thanks for all the work you have done jamesh. I wish I could loan you all of my free so you could implement the higher fps modes

Think I might start looking at those soon, trouble is they are going to take quite a while to sort out, and I actually have to do some real work at some point!

[slqa]

Hello,

I have a question about autofocus in this camera. Can I disable this option or set the fixed value of focus?

Sorry for english.