Calibrating your DVRs

User avatar
rangersac
Posts: 1394
Joined: Thu May 24, 2012 11:01 am
Location: Snug, Tasmania

Re: Calibrating your DVRs

Postby rangersac » Mon Aug 04, 2014 12:56 pm

Sticky please Mods.
De Rosa Macro | Intense Primer | Claud Butler Sovereign | Titus Motolite

User avatar
RonK
Posts: 9737
Joined: Mon Aug 24, 2009 2:08 pm
Location: Brisbane, Queensland
Contact:

Re: Calibrating your DVRs

Postby RonK » Mon Aug 04, 2014 5:46 pm

Quite astonishing that you are 1 metre wide.

The handlebars on the widest of my bikes is 60cm. I'm no sugar plum fairy but I'm definitely not wider.

How did you determine this dimension? Surely you risk credibility if it's not accurate?
Cycle touring blog and tour journals: whispering wheels...

User avatar
Mulger bill
Super Mod
Super Mod
Posts: 28906
Joined: Sun Sep 24, 2006 2:41 pm
Location: Sunbury Vic

Re: Calibrating your DVRs

Postby Mulger bill » Mon Aug 04, 2014 7:19 pm

rangersac wrote:Sticky please Mods.

Yep, done.
...whatever the road rules, self-preservation is the absolute priority for a cyclist when mixing it with motorised traffic.
London Boy 29/12/2011

maestro
Posts: 154
Joined: Wed Feb 29, 2012 1:06 pm

Video Camera Grids

Postby maestro » Fri Aug 22, 2014 9:16 pm

outnabike wrote:
maestro wrote:[quote=
When doing this you will need to be careful of the lean on your bike (ideally you should be able to find both ends of your measured line on the photo, as the line may not be exactly in line with the horizontal axis of the camera). Also, if you are travelling directly in line with the road (parallel to the kerb, not slightly diagonally) then you should be able to draw your centreline of travel with a line from the infinity point to your back wheel. This will need to be done on a per-frame basis as the camera/bike angle is not constant.


So with those assumptions, the car gave more than 1m clearance. If the lane was only 4m wide, then the green lines would be further out and the clearance may have been less than 1 metre.
Image

Apologies if I got over-technical.


My apologies to members for going off topic here I reckon this probably should have its own thread.

Sorry maestro and Oxford I got you both mixed up in my replies before.
Just to address that point maestro, and it is an interesting one. The more I study these grids the more ise the need to look at things from the camera's perspective.
so far I can see we are all a bit right and a bit wrong.
Perspective is in the eye of the beholder. We have to remember that the camera does not take in the view in a three dimensional sense as our eyes do. I did a layout to show the effect of the changing terrain behind the camera, remembering that the cam has been set up on level ground. And that is why the grid appears to dive down into the ground and when lifted (though not very well) in my above pic.
When the police come along and do a reconnaissance of a crash site they are well aware of these anomalies.
I have noted that it important to set the camera p with a mark on the wall for height as well as sideways view. Even without this you can tell how a camera is pointing by the horizon level in the frame.

Image

Here is my grid laid out on level ground but actually lifted to the correct camera eye level about 650mm off the ground.
Your wish to see the finish lines within the camera frame,(if I am getting your drift) can not happen as the periphery of the cameras view with that lens cant accommodate it within the frame and still maintain the grids width. Ok the centre of the grid is 750 mm and if I make that say 1 metre a bit more of the sides of the grid will disappear.
Image


OK, so I started a new thread to reply...

Firstly, yes, I was wrong when I concluded that your camera must have had a narrower field of view, the different grid could be purely because of camera angle.

With regard to your diagramme of the camera view on a hill... A consistent incline would have no affect on the grid lines as the bike angle (and therefore camera angle) would change by the same amount as the road changes. There would be issues with grid lines if the incline was not constant within the field of view (and also where the bike itself is) such as at the top of a crest, or bottom of a dip.

Your question about "finish lines" (I assume that you are referring to the convergence point)... This does not necessarily need to be within the camera field of view. There is space there, irrespective of whether you captured an image of it, and it does not make the theory any less valid if you haven't captured the convergence point in your image. There should be no reason why you cannot add a white area to the top of your frame so that you can draw the converging lines.

As far as raising your grid up to the camera height... Your gridlines will start higher, but must still converge at the same convergence point on the image, so raising the grid plane in space will result in the angles of your 'direction of travel' gridlines getting closer to being horizontal on the image. Now a single line directly out of your camera, parallel to the direction of travel, would go directly out from the lens to the convergence point and would appear as a single dot on the image. If you drew a grid at that height (i.e. the exact height of the camera) then all the lines of the grid would start at the height of the convergence point, and then converge... They would appear on the image as a single horizontal line through the convergence point. Also, a grid at any plane other than the plane of the road is useless as the point where a vehicle's wheels touch the road is really the only point where we can guarantee that we are talking about a point in the same plane as the grid (unless cars come with markings that show the point of the car at exactly your camera level, so we can match them with a grid at that level) so this is probably a moot point anyway.

There is also the issue of lens distortion (is that what you were trying to show with the curved black line in your image above?). You may also be able to calibrate that out with a video of a grid and correct for it using tools such as RawTherapee (if it is found to be significant).

Getting back to stock grids that you calibrate at home and superimpose without further calculations... You need to take into account any change in lean on the bike and change in how the bike sits on the suspension, compared to when the calibration was done, to see if there is any material error in the gridlines (assuming that the camera is mounted to the bike at the same place and angle). If you give up trying to catch drivers only giving you 80cm or more, then you can probably argue convincingly in court (with appropriate calculations addressing error margins due to lean and suspension) that there is 100% certainty that the gap was less than 1 metre.

User avatar
bychosis
Posts: 5065
Joined: Sat Jan 14, 2012 1:10 pm
Location: Lake Macquarie

Re: Video Camera Grids

Postby bychosis » Sat Aug 23, 2014 7:36 am

The variations in the camera location etc are the reason I prefer to have somewhere set I can calibrate my camera each ride. When I Leave the driveway I now always wheel the bike up the same line and therefore have a reference using the feature lines in the concrete. If necessary I will measure them then create a new overlay. One the way home from work I ride down the centre of a 3m path that has a centre line marking, again if necessary, eg the measuring tape out and create a new overlay. Haven't had to yet, but know that it can be done simply.
bychosis (bahy-koh-sis): A mental disorder of delusions indicating impaired contact with a reality of no bicycles.

User avatar
InTheWoods
Posts: 1841
Joined: Sun Jul 31, 2011 2:34 pm
Location: Brisbane

Re: Video Camera Grids

Postby InTheWoods » Sat Aug 23, 2014 6:29 pm

Mine:

Note that the lines show distance from my handlebars *on the ground*. So if the car's tyre is on the 75cm line, its body and mirror eat 15cm into that. The black line shows where tyres need to be to allow 10cm from tyre to tip of mirror.


User avatar
InTheWoods
Posts: 1841
Joined: Sun Jul 31, 2011 2:34 pm
Location: Brisbane

Re: Calibrating your DVRs

Postby InTheWoods » Sat Aug 23, 2014 6:30 pm

Another example, this one's for a mobius.

Note that the lines show distance from my handlebars *on the ground*. So if the car's tyre is on the 75cm line, its body and mirror eat 15cm into that. The black line shows where tyres need to be to allow 10cm from tyre to tip of mirror.


User avatar
bychosis
Posts: 5065
Joined: Sat Jan 14, 2012 1:10 pm
Location: Lake Macquarie

Re: Video Camera Grids

Postby bychosis » Sat Aug 23, 2014 6:43 pm

Looks great. the horizon point seems a little off though, it appears to be floating. You could also argue the relevance of the lines that far out. Maybe truncating them at a reasonable distance back would not reduce their effectiveness and remove the floating point
bychosis (bahy-koh-sis): A mental disorder of delusions indicating impaired contact with a reality of no bicycles.

Rhubarb
Posts: 949
Joined: Sat Jan 16, 2010 9:42 pm

Re: Calibrating your DVRs

Postby Rhubarb » Sat Aug 23, 2014 6:53 pm

Seriously good work guys (again).

User avatar
bychosis
Posts: 5065
Joined: Sat Jan 14, 2012 1:10 pm
Location: Lake Macquarie

Re: Video Camera Grids

Postby bychosis » Sat Aug 23, 2014 8:36 pm

This image of mine is of a straight, flat sharepath that is 3m wide. It indicates that a point is not really relevant as the far distance still doesn't meet, and to continue the lines would be above the horizon. The lines extend about three car spaces, probably about 6m each, but I haven't measured them (yet). Also of more is that a straight line doesn't quite fit the edge of the path in the foreground. It is only a bit out, but the side lines I have added are actually in two straight segments.

Image

The image is where I daily re-calibrate my camera on the way home. Just down the road from work, easy to ride the centreline for 20-50m to get a good image.
bychosis (bahy-koh-sis): A mental disorder of delusions indicating impaired contact with a reality of no bicycles.

User avatar
outnabike
Posts: 2081
Joined: Tue Jan 08, 2013 2:53 pm
Location: Melbourne Vic

Re: Video Camera Grids

Postby outnabike » Sat Aug 23, 2014 9:32 pm

maestro wrote:
outnabike wrote:
maestro wrote:[quote=
When doing this you will need to be careful of the lean on your bike (ideally you should be able to find both ends of your measured line on the photo, as the line may not be exactly in line with the horizontal axis of the camera). Also, if you are travelling directly in line with the road (parallel to the kerb, not slightly diagonally) then you should be able to draw your centreline of travel with a line from the infinity point to your back wheel. This will need to be done on a per-frame basis as the camera/bike angle is not constant.


So with those assumptions, the car gave more than 1m clearance. If the lane was only 4m wide, then the green lines would be further out and the clearance may have been less than 1 metre.
Image

Apologies if I got over-technical.


My apologies to members for going off topic here I reckon this probably should have its own thread.

Sorry maestro and Oxford I got you both mixed up in my replies before.
Just to address that point maestro, and it is an interesting one. The more I study these grids the more ise the need to look at things from the camera's perspective.
so far I can see we are all a bit right and a bit wrong.
Perspective is in the eye of the beholder. We have to remember that the camera does not take in the view in a three dimensional sense as our eyes do. I did a layout to show the effect of the changing terrain behind the camera, remembering that the cam has been set up on level ground. And that is why the grid appears to dive down into the ground and when lifted (though not very well) in my above pic.
When the police come along and do a reconnaissance of a crash site they are well aware of these anomalies.
I have noted that it important to set the camera p with a mark on the wall for height as well as sideways view. Even without this you can tell how a camera is pointing by the horizon level in the frame.

Image

Here is my grid laid out on level ground but actually lifted to the correct camera eye level about 650mm off the ground.
Your wish to see the finish lines within the camera frame,(if I am getting your drift) can not happen as the periphery of the cameras view with that lens cant accommodate it within the frame and still maintain the grids width. Ok the centre of the grid is 750 mm and if I make that say 1 metre a bit more of the sides of the grid will disappear.
Image


OK, so I started a new thread to reply...

Firstly, yes, I was wrong when I concluded that your camera must have had a narrower field of view, the different grid could be purely because of camera angle.

With regard to your diagramme of the camera view on a hill... A consistent incline would have no affect on the grid lines as the bike angle (and therefore camera angle) would change by the same amount as the road changes. There would be issues with grid lines if the incline was not constant within the field of view (and also where the bike itself is) such as at the top of a crest, or bottom of a dip.

Your question about "finish lines" (I assume that you are referring to the convergence point)... This does not necessarily need to be within the camera field of view. There is space there, irrespective of whether you captured an image of it, and it does not make the theory any less valid if you haven't captured the convergence point in your image. There should be no reason why you cannot add a white area to the top of your frame so that you can draw the converging lines.

As far as raising your grid up to the camera height... Your gridlines will start higher, but must still converge at the same convergence point on the image, so raising the grid plane in space will result in the angles of your 'direction of travel' gridlines getting closer to being horizontal on the image. Now a single line directly out of your camera, parallel to the direction of travel, would go directly out from the lens to the convergence point and would appear as a single dot on the image. If you drew a grid at that height (i.e. the exact height of the camera) then all the lines of the grid would start at the height of the convergence point, and then converge... They would appear on the image as a single horizontal line through the convergence point. Also, a grid at any plane other than the plane of the road is useless as the point where a vehicle's wheels touch the road is really the only point where we can guarantee that we are talking about a point in the same plane as the grid (unless cars come with markings that show the point of the car at exactly your camera level, so we can match them with a grid at that level) so this is probably a moot point anyway.

There is also the issue of lens distortion (is that what you were trying to show with the curved black line in your image above?). You may also be able to calibrate that out with a video of a grid and correct for it using tools such as RawTherapee (if it is found to be significant).

Getting back to stock grids that you calibrate at home and superimpose without further calculations... You need to take into account any change in lean on the bike and change in how the bike sits on the suspension, compared to when the calibration was done, to see if there is any material error in the gridlines (assuming that the camera is mounted to the bike at the same place and angle). If you give up trying to catch drivers only giving you 80cm or more, then you can probably argue convincingly in court (with appropriate calculations addressing error margins due to lean and suspension) that there is 100% certainty that the gap was less than 1 metre.



Quite right as the bike is on the same plane, this is what I meant to say.
Image

Lifting the grid on a bike that has the camera pointing down brings it back up to road view but then skews the grid and appears to make it longer. It also places the grid far point out of perspective. But I was just trying to show the effect of a downward facing camera. Doing a grid set out as I have done was for the purpose of getting a positive view of the road width as well as the cars distance.
You can get an idea of your width from the gutter as well.
But it is also to prove whether a camera is aligned correctly in two planes horizontal and vertical.
I suppose a straight line is simpler in the end though as it takes one plane "the vertical" out of concern.
I have not considered suspension at all as I am looking at road bikes, which most commuters would ride. But I know what you mean.Trying to keep it simple.
Vivente World Randonneur complete with panniers

User avatar
outnabike
Posts: 2081
Joined: Tue Jan 08, 2013 2:53 pm
Location: Melbourne Vic

Re: Video Camera Grids

Postby outnabike » Sat Aug 23, 2014 9:52 pm

Its good to get the different input and also I can rob some else's ideas. He He.
Please write into your posts which camera you are using as well. Then others can simply download it if they have the same camera to at least get a rough view of their set up.

It might be good to know....

if

Is the camera central
How far off centre it is.
Brand of camera
What settings it is on (eg 140, 120 degrees or even 1080 or 720 resolution. Just for the benefit of others and if you personally make a change.)


I think you have started a movement Oxford..... :)
Vivente World Randonneur complete with panniers

User avatar
Mulger bill
Super Mod
Super Mod
Posts: 28906
Joined: Sun Sep 24, 2006 2:41 pm
Location: Sunbury Vic

Re: Video Camera Grids

Postby Mulger bill » Sat Aug 23, 2014 11:13 pm

I'm really impressed by the efforts being gone to here and the willingness shown by members to share their intellectual property for the good of all.

Down the trail a ways, once the idea and implementation is a little more mature, I'd like to see some collaboration from the clever people here in creating a structured learning resource on creating, calibrating and editing the grids that can be stickied in Safety. What say you good members?

Shaun
...whatever the road rules, self-preservation is the absolute priority for a cyclist when mixing it with motorised traffic.
London Boy 29/12/2011

User avatar
bychosis
Posts: 5065
Joined: Sat Jan 14, 2012 1:10 pm
Location: Lake Macquarie

Re: Video Camera Grids

Postby bychosis » Sun Aug 24, 2014 7:46 am

Tend to agree, providing stock that is calibrated to one setup for others could cause more issues and reduce the overall effectiveness of a self calibrated camera. Imagine going to the police with a report with a well calibrated camera and them turning around and saying we checked the last guys setup and it was out by half a metre. Wastes time for all parties. These calibrations will probably not stand up in court especially if it was done by another person with another camera. I'm all for calibrating though.
bychosis (bahy-koh-sis): A mental disorder of delusions indicating impaired contact with a reality of no bicycles.

User avatar
InTheWoods
Posts: 1841
Joined: Sun Jul 31, 2011 2:34 pm
Location: Brisbane

Re: Video Camera Grids

Postby InTheWoods » Mon Aug 25, 2014 1:04 pm

bychosis wrote:Looks great. the horizon point seems a little off though, it appears to be floating. You could also argue the relevance of the lines that far out. Maybe truncating them at a reasonable distance back would not reduce their effectiveness and remove the floating point


The horizon point is off because I'm about to crest a hill :) It will naturally point up and into open air.

The original reason for sending the lines that far back was just to show they were parallel on the ground and all met back at the infinity point. I agree they have no relevance otherwise, apart from the "centreline" one, which on a straight road if I am riding in a continuous road position, and if everything is correct, should point to the path my tyres were following on the road. If they are pointing to the footpath or the other lane, then there's something wrong :)

I'll remove the >7m lines apart from the centreline from any future ones as they are kind of distracting. But if I do the thing that shows the process of calibration (scene with chalk lines, scene with chalk lines & photoshopped lines on top of chalk lines, just the lines without the background scene), I'll use the image with the lines going to infinity before reverting to the shorter version to overlay on the actual video.

Its handy to keep it all in a single image file (.psd if using photoshop) using different layers as you can turn on and off the layers you want and save a new .jpeg. I'll share at some point a comparison of the lines when the bike is straight up vs slightly tilted over.

User avatar
RedlandSteve
Posts: 99
Joined: Mon Feb 25, 2013 11:56 am

Re: Calibrating your DVRs

Postby RedlandSteve » Mon Aug 25, 2014 4:00 pm

Free image editor alternatives:

GIMP - I do not find this intuitive but worked it out in the end, very powerful on-par with photoshop.

Pixlr online editor - I like this one it is accessible online from the web browser.

User avatar
outnabike
Posts: 2081
Joined: Tue Jan 08, 2013 2:53 pm
Location: Melbourne Vic

Re: Calibrating your DVRs

Postby outnabike » Mon Aug 25, 2014 8:07 pm

I have never used Photoshop Elements, which is a lot cheaper. If any one has, does it feature a layering system and working in Photoshop native format?

The beauty of using Photoshop is that it is lossless in PSD, you save with all your layers and it is always there for recall. And from the PSD format you can save it as a GIF or JPEG.

I like the lines by some members that are not actually made opaque, as it takes another step out of the work and is easy to see. But I reckon the fewer the lines the better.
I am also thinking of doing a grid in three separate angles of a hill that can be tried on differing shots by individuals to see which suit the angle of their own camera set out.

Also when I get my Modius and have to set it up it may be good to get into some nice level place to set out my grid (on a nice Sunday) and invite some members with different cameras to drop in and we do some shots. Looks as though the Fly6 should get a look in and may be a Go pro. That would be a start to a bit of a collection of reasonable data. You could keep your cards for home and I could maybe download the info to my laptop. All results made available to any cyclist.
Any way just making a suggestion here. :)
Vivente World Randonneur complete with panniers

maestro
Posts: 154
Joined: Wed Feb 29, 2012 1:06 pm

Re: Calibrating your DVRs

Postby maestro » Mon Aug 25, 2014 9:11 pm

Oxford wrote:Yeah, after I posted, I did a quick search and found GIMP, you're right, not too intuitive but exactly what I am after. Just re-layering my pictures now so I can fix up the Google doc.


I used GIMP when I put lines on JasonC's image in the moron motorist's thread.

User avatar
Summernight
Posts: 2073
Joined: Mon Oct 08, 2012 3:40 pm
Location: Melbourne

Re: Calibrating your DVRs

Postby Summernight » Tue Aug 26, 2014 9:40 am

outnabike wrote:I have never used Photoshop Elements, which is a lot cheaper. If any one has, does it feature a layering system and working in Photoshop native format?


Yes, it does. The version of PS Elements I've used had layering. :D

User avatar
InTheWoods
Posts: 1841
Joined: Sun Jul 31, 2011 2:34 pm
Location: Brisbane

Re: Calibrating your DVRs

Postby InTheWoods » Tue Aug 26, 2014 1:10 pm

Oxford wrote:ITW, is there a free or cheap alternative to PhotoShop around? IrfanView can open PSD files but not save to them. I'd like to be able to do layered files, but like many people, not going to buy something like PhotoShop just for this purpose.


Microsoft has (had?) one called paint.net which seemed quite reasonable to me when I tried it. Had layers and everything.

User avatar
sumgy
Posts: 1487
Joined: Tue Mar 02, 2010 10:58 pm
Contact:

Re: Calibrating your DVRs

Postby sumgy » Wed Aug 27, 2014 9:19 am

I asked this on SCA's page but it was either missed or ignored.
Has anyone actually taken this method to QLD Police for their assessment of its validity?

User avatar
sumgy
Posts: 1487
Joined: Tue Mar 02, 2010 10:58 pm
Contact:

Re: Calibrating your DVRs

Postby sumgy » Wed Aug 27, 2014 10:00 am

Thanks for the reply.
I guess that time will tell then as to whether the overlay offers any additional benefits..

User avatar
outnabike
Posts: 2081
Joined: Tue Jan 08, 2013 2:53 pm
Location: Melbourne Vic

Re: Calibrating your DVRs

Postby outnabike » Wed Aug 27, 2014 10:24 am

Oxford wrote:
sumgy wrote:I asked this on SCA's page but it was either missed or ignored.
Has anyone actually taken this method to QLD Police for their assessment of its validity?

I have a number of complaints with QPS presently using this method/system. QPS requested unedited footage in addition to the overlayed footage for use in court if the matter goes that far. So the answer to your question is YES. I also do not use a form, I use the letter format that I have offered to Cycle. Forms are cold and impersonal and allow little latitude, letters are not, they allow you to convey emotion to let the reader know why the matter you are bringing to their attention is important to you. Forms can be ignored more easily, letters cannot.


Hi Oxford,
Same with me on my one and only current complaint to police as per my "punishment pass" thread.
I used your form letter and adapted it to my circumstance, sending it direct to the commissioner of police. I also made a series of full page stills from the video showing what I thought was relevant like the redgo plates, the empty lane next door, the drivers action within his car, the grid I had created on the ground and its relevance to the pics. Some pics had the grid on and off them to show the differing out looks. In short I didn't just present a video but a defence of my complaint and an allegation of careless driving on the part of the driver.

I received a call very quickly in about a week. The police man involved introduced himself and asked for a complete video of the unedited ride as they want to ascertain if the video is genuine and uncut.
The video is 15 minutes long, on a DVD and I noted the incident on the countdown clock for viewing.
I not that you have said elsewhere they may want a minute each side of the incident, and I am thinking they may want to turn the tables on me to find a fault with my riding or general law abiding. I love the bit where I cross a three lane road and dong a right hand turn showing a hand signal with a shadow on the ground as I ride.Talk about holier than though, I know I will go to heaven just on this video alone.... :)

Not so good is the bit where I ride on a path to get over a steep double lane truck infested overpass, as I would only be able to do 7 kph over it and there is no bike lane. So one foot back in hell... :)
Vivente World Randonneur complete with panniers

User avatar
InTheWoods
Posts: 1841
Joined: Sun Jul 31, 2011 2:34 pm
Location: Brisbane

Re: Calibrating your DVRs

Postby InTheWoods » Wed Aug 27, 2014 10:32 am

sumgy wrote:I asked this on SCA's page but it was either missed or ignored.
Has anyone actually taken this method to QLD Police for their assessment of its validity?


It doesn't need to be endorsed as "valid". Its simply an extra tool to assist with showing that a car passed less than 1m. In a court its a combination of your testimony and the camera footage which forms the evidence. If its simply to get police action, it can be useful when writing letters to assistant commissioners, police ministers and your local MP with a please explain :) Not that I've had to do that ever since my first brush off ... they've been good since then :)

The "calibration" doesn't pretend to be super accurate because its not (especially for almost-but-not-quite-1m-passes). I don't report anything in the >75cm range. I'd only bother with circa 65-75cm if there were aggravating circumstances like tailgating or horn blowing (offenses themselves). By the time you get to sub 65cm, there can be no doubt, with the assistance of the calibration lines, that the driver passed less than 1m away. Maybe it was 40cm, maybe it was 80cm, they're all well less than 1m. I have found the calibration lines useful myself to compare many passes that are borderline, so that I can better estimate passes in real life.

User avatar
outnabike
Posts: 2081
Joined: Tue Jan 08, 2013 2:53 pm
Location: Melbourne Vic

Re: Calibrating your DVRs

Postby outnabike » Thu Aug 28, 2014 7:51 pm

All this just as a matter of interest.
I have looked at our grids and lines and thought I would show the reasons for getting a camera correct in its alignment. We all want the camera to represent the true picture in case of your video being ultimately an evidential tool or as close to correct to at least hold up a valid point of view, which is probably as far as it will go.
Apart from that, some folk may wonder at the small amount of adjustment that can either fix or foul a good bit of proof of wrong doing.
This is a side on view of what the camera sees depending on its positioning. Nothing complex here but we get a different concept looking down the camera path when we see what the camera sees.
Nothing scientificley perfect but it illustrates the point.
Image

Firstly, a correctly calibrated camera. It shows the grid in relation to the road markings, all widths, and the gutter-line. It all combines to show the riders position on the road.
Image


Here the camera is tilted down 6 deg or 841mm. It gives a for-shortened view from the cameras perspective. The camera wants to see in a straight line. So overlaying the same correctly calibrated grid will give a distorted view. The blue line in all pics is the correct ground level
Image


The reverse happens with the camera tilted up 6 deg or 841 mm from ground level
Image


If we get that right there is the problem of misalignment left or right. Here the camera is 6 deg or 841 mm left to show the affect in the same pic. It makes the car appear closer and is not within the roads boundaries...................................................This is my calibrate correct grid but skewed left.
Image
Vivente World Randonneur complete with panniers

Who is online

Users browsing this forum: andrewjcw, Cyclophiliac, madmacca