How to time visual stimuli correctly

dvbridges

Hi there,

I am running some timing tests with Expyriment. I am using a LabJack to fire a TTL pulse and then measuring the time it takes for the stimuli to appear on screen (using a Black Box Tool Kit). Here is the minimal code - note, stimuli[0] is a blank screen, and stimuli[1] is a rectangle stimulus:

# Preload blank screen
trial.stimuli[0].preload()
expLabJack.writeRegister(6004, 0)
trial.stimuli[0].present(clear=False, update=False) 
# Preload stim
trial.stimuli[1].preload()
exp.clock.wait(300)

# Fire TTL pulse
expLabJack.writeRegister(6004, 1)
# present stim
trial.stimuli[1].present(clear=False, update=False)
exp.clock.wait(200)
expLabJack.writeRegister(6004, 0)

The problem I am having: Using this method, it takes approximately 28ms for the visual stimuli to be detected onscreen (visual stimuli onset - TTL pulse onset), so onset times of the stimuli are overshooting by at least a frame. Is there a better way of controlling the presentation of stimuli?

fladd

Hi,

in your example, nothing should be shown on the screen, since you explicitely set update=False.
What are the results of the visual presentation test in the test suite for you? Maybe we can go from there.

dvbridges

Thanks for the response. Apologies, that is a typo. The stimuli itself does not have an arg passed to the update parameter. Will get the test suite results now.

dvbridges

Here is the output of the test suite (had to cut some of the list entries due to character limit). If you would like it in a different format please let me know:

hardware_audio_card:
hardware_cpu_architecture: AMD64
hardware_cpu_details: Intel(R) Xeon(R) CPU E5-1620 v2 @ 3.70GHz
hardware_cpu_type: Intel64 Family 6 Model 62 Stepping 4, GenuineIntel
hardware_disk_space_free: 44015 MB
hardware_disk_space_total: 244196 MB
hardware_internet_connection: Yes
hardware_memory_free: 25342 MB
hardware_memory_total: 32683 MB
hardware_ports_parallel: []
hardware_ports_parallel_driver: None
hardware_ports_serial: []
hardware_video_card: RDPDD Chained DD
os_architecture: 64bit
os_details: SP1
os_name: Windows
os_platform: Windows
os_release: 7
os_version: 6.1.7601
python_expyriment_build_date: Thu Mar 9 13:48:59 2017 +0100
python_expyriment_revision: c4963ac
python_expyriment_version: 0.9.0
python_mediadecoder_version:
python_numpy_version:
python_pil_version:
python_pygame_version: 1.9.4
python_pyopengl_version: 3.1.0
python_pyparallel_version:
python_pyserial_version:
python_sounddevice_version:
python_version: 3.6.5
settings_folder: None
testsuite_visual_flipping_user: 1
testsuite_visual_opengl: 2
testsuite_visual_opengl_extensions: ['GL_AMD_multi_draw_indirect', 'GL_ARB_arrays_of_arrays', 'GL_ARB_base_instance', 'GL_ARB_blend_func_extended', 'GL_ARB_buffer_storage', 'GL_ARB_clear_buffer_object', 'GL_ARB_clear_texture', 'GL_ARB_color_buffer_float', 'GL_ARB_compatibility', 'GL_ARB_compressed_texture_pixel_storage', 'GL_ARB_conservative_depth', 'GL_ARB_compute_shader', 'GL_ARB_compute_variable_group_size', 'GL_ARB_copy_buffer', 'GL_ARB_copy_image', 'GL_ARB_debug_output', 'GL_ARB_depth_buffer_float', 'GL_ARB_depth_clamp', 'GL_ARB_depth_texture', 'GL_ARB_draw_buffers', 'GL_ARB_draw_buffers_blend', 'GL_ARB_draw_indirect', 'GL_ARB_draw_elements_base_vertex', 'GL_ARB_draw_instanced', 'GL_ARB_enhanced_layouts', ...[DELETED ENTRIES DUE TO CHAR LIMIT]]
testsuite_visual_opengl_renderer: b'NVS 310/PCIe/SSE2'
testsuite_visual_opengl_vendor: b'NVIDIA Corporation'
testsuite_visual_opengl_version: b'4.4.0'
testsuite_visual_pygame_driver: windib
testsuite_visual_pygame_screensize: (1920, 1080)
testsuite_visual_sync_refresh_rate: 60.0 Hz
testsuite_visual_timing_actual: [16.57459940179251, 33.29449532611761, 50.0559045758564, 16.66592866240535, 66.72127972706221, 33.18213266902603, 16.6681426926516, 33.33130379905924, 16.694434452801943, 33.220601675566286, 50.06171643617563, 16.662054069456644, 16.67976642784197, 33.33462486625649, 16.68281073216349, 49.99916972883511, 49.88210223382339, 66.64960009220522, 16.666482159052975, 16.66482162545435, 49.99833946931176, 66.6969252488343, ...]
testsuite_visual_timing_delay_histogram: {16: 5, 2: 9, 13: 13, 6: 8, 9: 12, 4: 9, 11: 12, 14: 11, 0: 12, 7: 12, 3: 9, 5: 9, 8: 12, 12: 12, 1: 10, 17: 1, 10: 13, 15: 11}
testsuite_visual_timing_delayed: 2.22 %
testsuite_visual_timing_inaccuracy: 0 ms
testsuite_visual_timing_todo: [0, 31, 37, 10, 57, 29, 5, 19, 17, 26, 47, 11, 8, 30, 5, 45, 36, 54, 9, 15, 41, 59, 25, 34, 33, 9, 28, 39, 44, 59, 48, 30, 11, 44, 54, 27, 2, 9, 43, 57, 18, 27, 8, 28, 26, 18, 7, 22, 20, 42, 31, 0, 34, 42, 39, 40, 4, 40, 1, 12, 52, 45, 22, 33, 52, 12, 50, 38, 21, 7, 16, 53, 5, 0, 46, 29, 7, 6, 40, 25, 23, 41, 19, 55, 56, 45, 17, 54, 56, 33, 49, 14, 2, 20, 12, 1, 37, 39, 4, 10, 6, 46, 35, 10, 46, 31, 13, 24, 2, 16, 53, 4, 11, 3, 20, 19, 48, 32, 49, 22, 27, 58, 32, 8, 59, 55, 17, 58, 14, 51, 24, 57, 3, 16, 43, 3, 34, 15, 6, 51, 23, 24, 36, 25, 37, 35, 1, 50, 56, 15, 13, 29, 52, 21, 53, 42, 48, 41, 51, 43, 30, 28, 47, 58, 50, 35, 55, 38, 36, 18, 47, 13, 23, 38, 26, 44, 49, 21, 14, 32]

dvbridges

The code without typo:

# Preload blank screen
trial.stimuli[0].preload()
expLabJack.writeRegister(6004, 0)
trial.stimuli[0].present(clear=False) 
# Preload stim
trial.stimuli[1].preload()
exp.clock.wait(300)

# Fire TTL pulse
expLabJack.writeRegister(6004, 1)
# present stim
trial.stimuli[1].present(clear=False)
exp.clock.wait(200)
expLabJack.writeRegister(6004, 0)

dvbridges

Ok, turns out the TTL should be fired after the call to present the stimuli. With that, I can achieve screen latencies of 11ms. Thanks for reading.