Buffers lenght are not equal using get

[LaurentBerger]

Hi,
In color Hue Saturation value on Iphone using a get with update mode = full (REST API) data lenght in same JSON answers buffer length for t, h ,v and s are not equal after N request. Value of is random.
In this screen copy numpy shape for buffer t, h s and v are printed. Everything is fined at the beginning (from 1 to 7 line) but for line 8 last buffer size 1149 and the other buffer 1148 :

{'build': '17272', 'depthFrontSensor': '1', 'fileFormat': '1.19', 'depthBackResolution': None, 'depthFrontResolution': None, 'depthBackRate': None, 'depthBackSensor': '1', 'deviceBaseOS': None, 'depthFrontRate': None, 'deviceBrand': 'Apple', 'deviceModel': 'iPhone13,4', 'version': '1.2.0', 'deviceManufacturer': None, 'deviceBoard': None, 'deviceCodename': None, 'deviceRelease': '18.6.1'}
{'crc32': '558d65ae', 'localCategory': 'Caméra', 'export': [{'set': 'Data', 'sources': [{'label': 't', 'buffer': 't'}, {'buffer': 'h', 'label': 'hue (°)'}, {'label': 'saturation', 'buffer': 's'}, {'label': 'value', 'buffer': 'v'}, {'label': 'Shutter Speed (s)', 'buffer': 'shutterSpeed'}, {'label': 'Aperture (f/N)', 'buffer': 'aperture'}, {'buffer': 'iso', 'label': 'ISO'}]}], 'category': 'Camera', 'localTitle': 'Color (Hue Saturation Value)', 'title': 'Color (Hue Saturation Value)', 'buffers': [{'name': 'lastPeriods', 'size': 0}, {'name': 'iso', 'size': 0}, {'name': 'framePeriod', 'size': 1}, {'name': 'frameRate', 'size': 1}, {'name': 'h', 'size': 0}, {'size': 0, 'name': 'shutterSpeed'}, {'size': 0, 'name': 'exposure'}, {'name': 'lastTimes', 'size': 5}, {'size': 0, 'name': 'v'}, {'size': 0, 'name': 'aperture'}, {'name': 's', 'size': 0}, {'size': 0, 'name': 't'}], 'inputs': []}
Shape :  (66,) (66,) (66,) (66,)
Shape :  (189,) (189,) (189,) (189,)
Shape :  (290,) (290,) (290,) (290,)
Shape :  (426,) (426,) (426,) (426,)
Shape :  (549,) (549,) (549,) (549,)
Shape :  (692,) (692,) (692,) (692,)
Shape :  (1010,) (1010,) (1010,) (1010,)
Shape :  (1148,) (1148,) (1148,) (1149,)

Full data with json answer for meta config and get

nouveau 2.txt

There is no problem using phyphox-android

[Staacks]

Sorry, I did not yet have time to test a few things, but since this has been open for a few days now, let me add my first thoughts:

Am I correct with the assumption that it is always the t buffer that has more entries thanm h, v and s? Also, I would assume that if you pause the experiment and use the export function you are getting the same data as you are getting from the REST API at that point, right?

Can you see if the data points are lost or if they are just delayed? i.e. when you pause the experiment do the buffers catch up or do they remain at differen sizes?

So, I think that the phone can not keep up with the data analysis for some reason with two possible consequences:

• It could just be that the REST request is asynchronous to the data analysis (you get the time stamp from the latest video frame before the GPU has finished analyzing it, so that data will just follow in the next request). That's not ideal but a minor issue.
• The analysis could fall behind so much that the frame is dropped and not analyzed at all and you still get the timestamp. That's definitely a bug and should also show in the data regardless of the REST API. If the phone cannot keep up, dropping a frame can eventually not be prevented, but in that case the timestamp should also be dropped.

The thing that is a bit weird is that you are using an iPhone 12 Pro, which should have plenty of headroom. Does the selected area for the data analysis make a difference, i.e. does the problem occur sooner if you select the entire image vs. just a tiny area?

I could also imagine that the REST API could cause the phone not to be able to keep up. You are requesting the entire dataset about twice per second. Getting 4000 values and assembling them into a ~80kB JSON string (it's 20B per value in JSON with the precision phyphox uses) is not unmanagable for a modern phone, but it's not nothing...

(And sorry, I did not get the part "Value of is random." Something seems to be missing here.)

[LaurentBerger]

"The value of N is random."
It's difficult to test because It is student's phone.
Problem laso occurs too with gyroscope

[LaurentBerger]

Now I have got more data.
Here shape of each buffer (retrieve with URL full) and data response retrieve for each response (file index 0 to 19)

reponse0_19.zip

<- 0 ->
acc_time	151
accX	151
accY	151
accZ	151
acc	151
<- 1 ->
acc_time	294
accX	294
accY	294
accZ	294
acc	294
<- 2 ->
acc_time	458
accX	457
accY	458
accZ	458
acc	458
<- 3 ->
acc_time	743
accX	743
accY	743
accZ	743
acc	743
<- 4 ->
acc_time	896
accX	895
accY	895
accZ	895
acc	896
<- 5 ->
acc_time	1292
accX	1291
accY	1292
accZ	1291
acc	1292
<- 6 ->
acc_time	1540
accX	1539
accY	1539
accZ	1539
acc	1539
<- 7 ->
acc_time	1753
accX	1753
accY	1753
accZ	1753
acc	1753
<- 8 ->
acc_time	2093
accX	2093
accY	2092
accZ	2093
acc	2093
<- 9 ->
acc_time	2243
accX	2243
accY	2243
accZ	2243
acc	2243
<- 10 ->
acc_time	2478
accX	2478
accY	2478
accZ	2478
acc	2478
<- 11 ->
acc_time	2621
accX	2621
accY	2620
accZ	2621
acc	2621
<- 12 ->
acc_time	2801
accX	2801
accY	2801
accZ	2801
acc	2801
<- 13 ->
acc_time	3886
accX	3886
accY	3886
accZ	3886
acc	3886
<- 14 ->
acc_time	4182
accX	4182
accY	4182
accZ	4182
acc	4181
<- 15 ->
acc_time	4520
accX	4520
accY	4521
accZ	4521
acc	4520
<- 16 ->
acc_time	4799
accX	4799
accY	4799
accZ	4799
acc	4798
<- 17 ->
acc_time	5297
accX	5297
accY	5297
accZ	5297
acc	5296
<- 18 ->
acc_time	5698
accX	5698
accY	5698
accZ	5698
acc	5697
<- 19 ->
acc_time	5988
accX	5988
accY	5988
accZ	5988
acc	5987

experiment config

crc32 : e7e65d83
title : Acceleration (without g)
localTitle : Accélération (sans g)
category : Raw Sensors
localCategory : Capteurs
buffers : [{'size': 0, 'name': 'acc_time'}, {'size': 0, 'name': 'accX'}, {'size': 0, 'name': 'acc'}, {'size': 0, 'name': 'accZ'}, {'name': 'accY', 'size': 0}]
inputs : [{'outputs': [{'x': 'accX'}, {'y': 'accY'}, {'z': 'accZ'}, {'abs': 'acc'}, {'t': 'acc_time'}], 'source': 'linear_acceleration'}]
export : [{'sources': [{'buffer': 'acc_time', 'label': 'Time (s)'}, {'buffer': 'accX', 'label': 'Linear Acceleration x (m/s^2)'}, {'label': 'Linear Acceleration y (m/s^2)', 'buffer': 'accY'}, {'buffer': 'accZ', 'label': 'Linear Acceleration z (m/s^2)'}, {'label': 'Absolute acceleration (m/s^2)', 'buffer': 'acc'}], 'set': 'Raw Data'}]

version : 1.2.0
build : 17272
fileFormat : 1.19
deviceModel : iPhone14,5
deviceBrand : Apple
deviceRelease : 18.6.2
depthFrontSensor : 1
depthBackSensor : 0