Re: Updates, autofocus, 5Mpix mode on N900? Re: [RFC 08/13] smiapp-pll: Take existing divisor into account in minimum divisor check

From: Sakari Ailus <sakari.ailus@iki.fi>
To: Pavel Machek <pavel@ucw.cz>
Cc: sre@kernel.org, pali.rohar@gmail.com,
	linux-media@vger.kernel.org, linux-kernel@vger.kernel.org,
	laurent.pinchart@ideasonboard.com, mchehab@kernel.org,
	ivo.g.dimitrov.75@gmail.com
Subject: Re: Updates, autofocus, 5Mpix mode on N900? Re: [RFC 08/13] smiapp-pll: Take existing divisor into account in minimum divisor check
Date: Thu, 23 Mar 2017 09:32:07 +0200	[thread overview]
Message-ID: <20170323073207.GO10701@valkosipuli.retiisi.org.uk> (raw)
In-Reply-To: <20170322234651.GB5831@amd>

Hi Pavel,

On Thu, Mar 23, 2017 at 12:46:51AM +0100, Pavel Machek wrote:
> On Tue 2017-02-28 16:16:21, Sakari Ailus wrote:
> > On Tue, Feb 28, 2017 at 03:09:21PM +0100, Pavel Machek wrote:
> > > Can I get you to apply this one? :-).
> > 
> > Let me try to understand again what does that change actually do. I'll find
> > the time during the rest of this week.
> > 
> > I'm starting to think we need a test suite for the PLL calculator...
> 
> Any update here or on the other patch? We are quite close to working
> camera now...

I've been working on PLL test cases. The PLL calculator really requires that
to be able to test any changes made to it.

> 
> Plus I have played with v4l-utils, and managed to implement autofocus
> and autoexposure -- it was easier than expected. I believe you
> mentioned you had some patches to automatically initialize the
> pipeline. Do you and can I have them?

It was an early prototype and it wasn't really functional yet.

Given a video node, it can find possible pipelines to the image sources with
common formats. I.e. the ccdc -> rsz path is not available for raw cameras.

C (especially without helper libraries) wasn't particularly suitable for the
task, the data structures I had didn't end up too nice. What would also be
necessary is to associate library or application specific data to entities,
this could be as simple as key-value pairs with both key and value being
pointers.

> 
> Last thing.. Is someone able to compute new modes for et8ek8? I
> believe smaller than 640x480 mode would be useful for video streaming,
> and I still can't get 5MPix mode to work; understanding what goes on
> there would be useful.

Unfortunately the et8ek8 does not conform to a standard such as SMIA. :-(
I'm not sure the datasheet provides enough information to come up with new
mode definitions. Perhaps with some experimentation it could be possible.
There are a few additional embedded documents in it, those are worth
checking out. LibreOffice can open them AFAIR.

-- 
Regards,

Sakari Ailus
e-mail: sakari.ailus@iki.fi	XMPP: sailus@retiisi.org.uk