Ball pitching speed / Shutter speed accuracy experiment

[K-amps]

I am trying to estimate the velocity of a pitcher (the radar gun I have does not seem to be not accurate) using a framed capture at 1/60th of a second.

I set a 1d2 at 1/60th sec capture then measured the trail of the ball and (how much it moved in 1/60th of a second); did some on screen measurements; ran some calculations; to get at an expected speed of 90.12mph.

I think this is very close to what I thought his speed would be, so I was confident in the modelling.

This was a cricket pitch, but lets use baseball analogy if it helps.

We have another pitcher that we think might be a bit faster or at least the same than Pitcher #1.

This time I set a 5Dclassic (that I had on me at the time) with the similar (though not same) framing and 1/60th frame capture. I figured as long as I can get a ratio of the ball trail vs the diameter of the ball, I’d end up with a constant that could be uses to compare the two captures and calculate the speeds. We also have a third pitcher who we feel is a bit slower than #1 and #2 that I framed the second time as well.

To summarize:

Pitcher #1 = 1d2 1/60 sec
Pitcher #2 = 5D 1/60 sec
Pitcher #3 = 5D 1/60 sec

For that batters (batsmen) who have faced these 3 guys feel Pitcher 1 & 2 and neck and neck at around 90mph, while pitcher 3, is about 7-8% slower.

Results:

Pitcher 1 = 90mph (1d2)
Pitcher 2 = 77mph (5d)
Pitcher 3 = 67mph (5d)

While the speed differential between pitcher 2 & 3 is absolute since the test and the conditions are identical (even the trail in the photo bears out a 14% difference between them) what I cannot believe is that Pitcher 2, is only 77mph (though he seems very close or faster than pitcher 1).

I may not be able to assemble these guys together for a retake with the same equipment so here is my question:

When the 1d2 takes a picture at 1/60th of a second, how accurate is that? Same for the 5D.

Based on this, it seems to me that there is a 15% difference between the 2 bodies. Should I trust one or the other? Are shutter speeds that accurate?

 

[dougkerr]

Hi, K,

I am trying to estimate the velocity of a pitcher (the radar gun I have does not seem to be not accurate) using a framed capture at 1/60th of a second.

Does the "scan" aspect of the FP shutter introduce any complication into your process? Perhaps with a shutter speed of 1/60 s it would not be a factor.

Is the travel of the ball parallel to the focal plane (or at a consistent angle to it)?

Interesting project.

Best regards,

Doug

 

[K-amps]

Hi, K,

I am trying to estimate the velocity of a pitcher (the radar gun I have does not seem to be not accurate) using a framed capture at 1/60th of a second.

Does the "scan" aspect of the FP shutter introduce any complication into your process? Perhaps with a shutter speed of 1/60 s it would not be a factor.

Is the travel of the ball parallel to the focal plane (or at a consistent angle to it)?

Interesting project.

Best regards,

Doug

Doug: thanks for reading through my "blog" there.

I am afraid I have to plead ignorance to the ""scan" aspect of the FP shutter" question. This was a series of stills, the capture should have lasted 1/60th of a second in my mind.

The angles and other aspects of the ball line were far from perfect, for Pitcher 1, the 1d2 was handheld and there is vertical blurring, but it has little impact (little) on the horizontal trajectory. A +/- 5% margin of error is very much a given; the way i did the experiment, but I am getting an over 10% difference from expected result.

See the attached pictures and see what I mean. The blur of the ball is clearly different lengths. Pitcher #2 should be faster or equal to Pitcher #1.

 

[acearchie]

From the pictures you have provided I think you will find that the problem lies that on the photo with the 1d2 you are not parallel to the line in which the ball will be traveling. This means that the ball will be traveling horizontally and vertically (downwards) as seen in the photo but also in a third 'z' direction towards the camera which would give the impression that it would have travelled less far.

Now I have written that, I have just thought that this would account for the 1d2 shot being slower but does not work if you think he is faster.

The other factor that may influence your results is that the first photo is taken earlier in the bowling action and therefore will have more horizontal velocity as it has just left the hand. The other two photo's are later in the bowling action and therefore the ball will have slowed down with the the air resistance and other variables. These may be minute differences, but if you are looking at getting your experiment right then you should try and take the photos at the same time of the bowling action.

Another thought that I have just had whilst writing is that each bowler might have had an exceptionally good bowl or bad bowl. #1 may have bowled his fastest ever whereas #2 bowled slower than average.

Just like science experiments at school you should have taken at least 3 shots of each bowler to ensure that you were left with an average speed rather than potentially an anomalous result.

Did you find the diameter of the ball and then find out how many ball lengths fit into the blur?

Were you measuring true horizontal velocity and vertical velocity?

Also if you do get to try it again whack the camera on the tripod as you seem to have to hand shake which doesn't help when measuring. Therefore you can round to the nearest mile an hour as the data you have is not accurate enough to be within 2 decimal places.

Hope this helped!

 

[Kernuak]

A couple of other thoughts, there could have been differing weather conditions, but also, in the first shot it looks like there was also horizontal camera movement. Looking at the stumps and the two other people in #1, it looks like the camera was moved from left to right. This was one of the inherent problems found with handheld speed radars, where cars were measured as travelling quicker than they actually were and why the cases where dismissed from court, prior to re-training.

 

[D n L Photography]

The question that I would have is "Does the difference in the crop factor make any difference?". The 1d2 has a 1.3 crop factor whereas the 5d is FF

 

[acearchie]

The question that I would have is "Does the difference in the crop factor make any difference?". The 1d2 has a 1.3 crop factor whereas the 5d is FF

This wouldn't be a problem as the ball is a constant size so if the OP has measured the distance using the diameter of the ball then this would be the same for each picture as the ball is a standard size.

 

[K-amps]

The question that I would have is "Does the difference in the crop factor make any difference?". The 1d2 has a 1.3 crop factor whereas the 5d is FF

This wouldn't be a problem as the ball is a constant size so if the OP has measured the distance using the diameter of the ball then this would be the same for each picture as the ball is a standard size.

Correct, the ball is 6.7cms and I am going off the ratios of width of the ball vs. length of the blur.

I did take 3-4 shots each, and these are the longest "blurs" for each bowler. The bowler's had relatively consistent blurs in the 3 shots I took.

There is some movement in shot #1 but it is more vertical (look at the blur of the ball, there a slight hump); without that movement, the ball would have been even faster, so that's not what is causing my 15% difference.

Goes back to the accuracy of the shutter speed. I was googling "shutter speed accuracy" and came about some experiments people did at 1/100th of a sec and found after 75 shots, the shutter was languishing at 1/140 -1/160th of a second... sadly I forget what camera it was.

So does anyone know if Canon shutter speeds are accurate or does +/- 15% fall under an "acceptable tolerance" ?

 

[acearchie]

So does anyone know if Canon shutter speeds are accurate or does +/- 15% fall under an "acceptable tolerance" ?

I think that the tolerance would be much higher as otherwise there would be a +/- 15% (therefore potentially 30%) difference in exposure in some pictures.

I also wonder how the shutter speed time varies as the shutter speeds get faster?

 

[DJL329]

Have you tried doing a similar test using an automobile, traveling at a predetermined speed? When you compare the results from the 2 bodies, the results should be identical (if they're both working properly), since the speed of the car is now a constant.

You could also do that to test your radar gun. If it is off by a certain factor (the percentage difference is the same, even at different speeds), you could then factor it in when using it on the pitchers. Or possibly (I don't know anything about radar guns) use it to calibrate the radar gun?

 

[hendrik-sg]

If you want to mesure the shutter speed of a cam i would recommend a more precise, rtepetable setup.

Maybe you have access to a industry electric motor which runs with 1-2 % accuracy synchronous to the elecric grid. with 50 or 60Hz depeding on where you live. if you can fix a disc on the motor you can mark the disc with a sign and take pictures with blur. in this way you should be able to mesure about 60 to 600 hz with better accuracy.