summaryrefslogtreecommitdiffstats
path: root/Documentation/usb/ov511.txt
blob: bc9cad2abb54d7922a3961b29db30965be73e9f4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
-------------------------------------------------------------------------------
Readme for Linux device driver for the OmniVision OV511 USB to camera bridge IC
-------------------------------------------------------------------------------

Author: Mark McClelland
Homepage: http://alpha.dyndns.org/ov511

INTRODUCTION:

This is a driver for the OV511, a USB-only chip used in many "webcam" devices.
Any camera using the OV511/OV511+ and the OV7610/20/20AE CCD should work. It 
supports streaming and capture of color or monochrome video via the Video4Linux
API. Most V4L apps are compatible with it, but a few video-conferencing programs
do not work yet. The following resolutions are supported: 640x480, 448x336,
384x288, 352x288, and 320x240.

If you need more information, please visit the OV511 homepage at the above URL.

WHAT YOU NEED:

- If you want to help with the development, get the chip's specification docs at
  http://www.ovt.com/omniusbp.html

- A Video4Linux compatible frame grabber program (I recommend vidcat and xawtv)
    vidcat is part of the w3cam package:  http://www.hdk-berlin.de/~rasca/w3cam/
    xawtv is available at:  http://www.in-berlin.de/User/kraxel/xawtv.html

HOW TO USE IT:

You must have first compiled USB support, support for your specific USB host
controller (UHCI or OHCI), and Video4Linux support for your kernel (I recommend
making them modules.)

Next, (as root) from your appropriate modules directory (lib/modules/2.3.XX):

	insmod usb/usbcore.o
	insmod usb/usb-uhci.o  <OR>  insmod usb/ohci-hcd.o
	insmod misc/videodev.o
	insmod usb/ov511.o

If it is not already there (it usually is), create the video device:

	mknod /dev/video c 81 0

Sometimes /dev/video is a symlink to /dev/video0

You will have to set permissions on this device to allow you to read/write
from it:

	chmod 666 /dev/video
	chmod 666 /dev/video0 (if necessary)
	
Now you are ready to run a video app! Both vidcat and xawtv work well for me
at 640x480.
	
[Using vidcat:]

	vidcat -s 640x480 > test.jpg
	xview test.jpg
	
[Using xawtv:]

You must make some modifications to the source and compile it before you use it.
(Note: this may not be applicable to versions other than 3.06)

In src/Xawtv.ad, change xawtv.tv.width to 640 and xawtv.tv.height to 480. Next,
in src/grab-v4l.c, change SYNC_TIMEOUT from 1 to 2. Then, from the main xawtv
directory:

	make clean
	./configure
	make
	make install

Now you should be able to run xawtv. Right click for the options dialog. If
you get a scrambled image it is likely that you made a mistake in Xawtv.ad.
Try setting the size to 320x240 if all else fails.

MODULE PARAMETERS:

  You can set these with:  insmod ov511 NAME=VALUE
  There is currently no way to set these on a per-camera basis.

  NAME: autoadjust
  TYPE: integer (boolean)
  DEFAULT: 1
  DESC: The camera normally adjusts exposure, gain, and hue automatically. This
        can be set to 0 to disable this automatic adjustment. Note that there is
        currently no way to set these parameters manually once autoadjust is
        disabled.

  NAME: debug
  TYPE: integer (0-6)
  DEFAULT: 3
  DESC: Sets the threshold for printing debug messages. The higher the value,
        the more is printed. The levels are cumulative, and are as follows:
          0=no debug messages
          1=init/detection/unload and other significant messages
          2=some warning messages
          3=config/control function calls
          4=most function calls and data parsing messages
          5=highly repetitive mesgs

  NAME: fix_rgb_offset
  TYPE: integer (boolean)
  DEFAULT: 0
  DESC: Some people have reported that the blue component of the image is one
        or so lines higher than the red component. This is only apparent in 
        images with white objects on black backgrounds at 640x480. Setting this
        to 1 will realign the color planes correctly. NOTE: This is still
        experimental and very buggy. You will likely need a fast (500 MHz) CPU.

  NAME: snapshot
  TYPE: integer (boolean)
  DEFAULT: 0
  DESC: Set to 1 to enable snapshot mode. read() will block until the snapshot
        button is pressed. Note that this does not yet work with most apps,
        including xawtv and vidcat. NOTE: See the section "TODO" for more info.

  NAME: sensor
  TYPE: integer ([0, 1, 3])
  DEFAULT: [varies]
  DESC: If you know that your camera sensor is not detected correctly, set this
        parameter. This is a global option for all attached OV511 cameras. You
        will probably never need to set this, but if you do, valid values are:
        	0 for OV7620
        	1 for OV7620AE
        	3 for OV7610

  NAME: i2c_detect_tries
  TYPE: integer (don't set it insanely high!)
  DEFAULT: 5
  DESC: This is the number of times the driver will try to sync and detect the
        internal i2c bus (which connects the OV511 and sensor). If you are
        getting intermittent detection failures ("Failed to read sensor ID...")
        you should increase this by a modest amount. If setting it to 20 or so
        doesn't fix things, look elsewhere for the cause of the problem.

  NAME: aperture
  TYPE: integer (0 - 15)
  DEFAULT: [varies by sensor]
  DESC: For legal values, see the OV7610/7620 specs under register Common F.
        This setting affects the upper nybble of that reg (bits 4-7). This is
        for if you want to play with the camera's pixel saturation.

  NAME: force_rgb
  TYPE: integer (boolean)
  DEFAULT: 0
  DESC: Force image to be read in RGB instead of BGR. This option allow
        programs that expect RGB data (e.g. gqcam) to work with this driver. If
        your colors look VERY wrong, you may want to change this.

  NAME: buf_timeout
  TYPE: integer
  DEFAULT: 5 (seconds)
  DESC: Number of seconds before unused frame buffers are deallocated.
        Previously, memory was allocated upon open() and deallocated upon
        close(). Deallocation now occurs only if the driver is closed and this
        timeout is reached. If you are capturing frames less frequently than
        the default timeout, increase this. This will not make any difference
        with programs that capture multiple frames during an open/close cycle.

  NAME: cams
  TYPE: integer (1-4 for OV511, 1-31 for OV511+)
  DEFAULT: 1
  DESC: Number of cameras allowed to stream simultaneously on a single bus.
        Values higher than 1 reduce the data rate of each camera, allowing two
        or more to be used at once. If you have a complicated setup involving
        both OV511 and OV511+ cameras, trial-and-error may be necessary for
        finding the optimum setting.

  NAME: retry_sync
  TYPE: boolean
  DEFAULT: 0
  DESC: Prevent apps from timing out if frame is not done in time. This is
        useful if you are having problems with Xawtv getting "stuck" on a frame
        when your system is under heavy load.

  NAME: sensor_gbr
  TYPE: boolean
  DEFAULT: 0
  DESC: This makes the sensor output GBR422 instead of YUV420. This saves the
        driver the trouble of converting YUV to RGB, but it currently does not
        work very well (the colors are not quite right)

WORKING FEATURES:
 o Color streaming/capture at 640x480, 448x336, 384x288, 352x288, and 320x240
 o RGB24, RGB565, YUV420, YUV422, YUYV, and YUV422P color
 o Monochrome
 o Setting/getting of saturation, contrast, brightness, and hue (only some of
   them work the OV7620 and OV7620AE)
 o /proc status reporting

EXPERIMENTAL FEATURES:
 o fix_rgb_offset: Sometimes works, but other times causes errors with xawtv and
   corrupted frames. If you have a very fast CPU, you can try it.
 o Snapshot mode (only works with some read() based apps; see below for more)
 o OV6620 sensor support
 o GBR422 parsing
 o 160x120

TODO:
 o Fix the noise / grainy image problem.
 o Get compression working. It would be a nice addition as it improves
   frame rate quite a bit. OmniVision wouldn't tell me how the algorithm works,
   so we can't really work on that yet. Please kindly inform OmniVision that you
   would like them to release their specifications to the Linux community.
 o YUV422
 o Fix fixFrameRGBoffset(). It is not stable yet with streaming video.
 o V4L2 support (Probably not until it goes into the kernel)
 o Get rid of the memory management functions (put them in videodev.c??)
 o Setting of contrast and brightness not working with 7620/7620AE
 o Driver/camera state save/restore for when USB supports suspend/resume
 o Unstable on SMP systems
 o OV7620/OV6620 experience frame corruption with moving objects
 o OV6620 is too dark
 o 176x144 support
 o Driver sometimes hangs upon close() with OHCI
 o The image should always be written properly to the mmap'ed buffer as long as
   the requested image size is at least the minimum size. This will likely
   require a rewrite of all the parsing code.

HOW TO CONTACT ME:

You can email me at mwm@i.am . Please prefix the subject line
with "OV511: " so that I am certain to notice your message.

CREDITS:

The code is based in no small part on the CPiA driver by Johannes Erdfelt,
Randy Dunlap, and others. Big thanks to them for their pioneering work on that
and the USB stack. Thanks to Bret Wallach for getting camera reg IO, ISOC, and
image capture working. Thanks to Orion Sky Lawlor, Kevin Moore, and Claudio
Matsuoka for their work as well.