How are the data of time stamp and frame measured?

[imarin18]

Dear Sir or Madam,

I measured my eye movements at the sampling frequency of 120Hz on Unity platform. I recorded the data of time stamp, frame, and other eye parameters such as pupil diameter. The following is a part of my measurement data.

time_stamp(us)       frame      pupil_diameter_L(mm)    pupil_diameter_R(mm)
2515185                   46129     4.857666                            4.549805
2515185                   46129     4.847290                            4.549103

As you can see, while the time stamp and frame have the same values in two lines, the pupil diameter have different values. I guess that the eye data are properly measured at 120Hz samping frequency, whereas the data of time stamp and frame are probably read by another way (e.g. reading time and frame information from laptop or Unity time function). Could anyone please comment on if my understanding is correct?

 

Best regards,

Imarin
@Daniel_Y @zzy

[Daniel_Y]

Which SR_runtime version are you using?

[imarin18]

Dear @Daniel_Y,

Thank you very much for your reply.

I use SRanipal SDK v1.1.0.1.

 

I actually investigated further after my first inquiry. I understood that it would be possible to see the same frame sequence because the frame rate would be lower than eye tracking sampling rate. However, I do not know why the time stamp did not change.

 

Speaking of sampling frequency, I observed that the sampling might not be stable and be lower than 120Hz for the initial 30 seconds, when I measured my eye movements on Unity for 2 minutes. The attached figure shows the sampling interval in milliseconds between each sample. As you can see, the interval for some inital samples is quite longer than the expected sampling interval (1/120Hz = 8.3 ms) and then the interval becomes stable at around 8.3 ms afterwards. I guess that some initialisation process is running for the initial 30 seconds, causing the larger intervals. Do you have any thoughts about the condition?

 

Best regards,

Imarin 

 

[Daniel_Y]

Is SR runtime 1.1.2.0?

[imarin18]

Dear Daniel,

Yes the version is 1.1.2.0.

Best regards,

Imarin

 

@Daniel_Y

[Daniel_Y]

On 11/14/2019 at 12:53 AM, imarin18 said:

Dear Sir or Madam,

I measured my eye movements at the sampling frequency of 120Hz on Unity platform. I recorded the data of time stamp, frame, and other eye parameters such as pupil diameter. The following is a part of my measurement data.

time_stamp(us)       frame      pupil_diameter_L(mm)    pupil_diameter_R(mm)
2515185                   46129     4.857666                            4.549805
2515185                   46129     4.847290                            4.549103

As you can see, while the time stamp and frame have the same values in two lines, the pupil diameter have different values. I guess that the eye data are properly measured at 120Hz samping frequency, whereas the data of time stamp and frame are probably read by another way (e.g. reading time and frame information from laptop or Unity time function). Could anyone please comment on if my understanding is correct?

 

Best regards,

Imarin
@Daniel_Y @zzy

Thanks for your feedback. We noticed this issue and will be addressed in the next release.

[imarin18]

Dear Daniel,

 

Thank you very much for your reply. I understood.

May I understand that eyes are sampled exactly at 120Hz? Or is it possible to see that the sampling frequency of eye tracking device changes? For instance, the sampling frequency is sometimes 120Hz but sometimes it changes to 130Hz, 70Hz, etc.

 

Best regards,

Imarin

 

@Daniel_Y

[imarin18]

Dear @Daniel_Y,

 

I checked the data of timestamp with SR runtime 1.1.2.0 version. I observed that the value of time stamp was consistent for around 1.7 seconds after starting the measurement of eye movements. Then, it seemed that the time stamp was recorded properly. However, as you can see the attached graph, the sampling interval is not stable. It is understandable to see the interval fluctuates between 8 and 9 milliseconds, but the interval often goes up to around 16 milliseconds.

Does it mean that the sampling is sometimes not performed exactly at 120Hz rate? Or is it because of the latency of calculation on computer?

 

 

Thank you very much for your feedback you could provide.

 

Best regards,

Imarin

[wenjiec_shen]

Hi @imarin18,

In order confirm your state, could you please provide more information?

For example, your computer equipment, the meaning of x axis in your chart and how you get the timestamp in source code.

Thank you.

Edited December 17, 2019 by wenjiec_shen

[imarin18]

Dear @wenjiec_shen

Thank you very much for your response. I try to measure eye movements with using Unity. The figure above has the sampling point on x-axis and the sampling interval on y-axis in milliseconds. The sampling interval was calculated by ViveSR.anipal.Eye.EyeData_v2.timestamp that was recorded by VIVE Pro eye tracking device. However, as @Daniel_Y acknowledged above, the time stamp information was not correct, wasn't it? Thus, the graph I put above was also not correct. I'm sorry that I just now noticed this.

I hope that the issue of time stamp will be resolved soon in the next release.

Best regards,

@imarin18