My Canon R3 sensor readout speed measurements

A few months ago I started an open source / crowd source project on GitHub to create a definitive repository of sensor readout speed measurements, for both stills and video. The project uses the built-in LED on an inexpensive Arduino board as a standardized light source with custom software that toggles the LED at a known fixed frequency (500Hz). This assures all the measurements are calibrated and accurate to the same source.

I just published my results for the R3, which can be viewed at the link below. All of the columns in the live table are sortable by clicking on the column header - for example you can sort by photographic readout speed, 4K30p readout speed, etc... You can also toggle between milliseconds and 1/sec displays of the results. And also filter the results by entering a search pattern.

Canon R3 Sensor Readout Measurements

More detailed R3 measurements:

Canon R3 Sensor Readout Measurements (detail view)

For those interested here is the GitHub homepage of my project - it has lots of foundational information about rolling shutters, along with more details on the methodology employed for the measurements:

https://github.com/horshack-dpreview/RollingShutter

But what do your pictures look like?

KiloHotelphoto said:

But what do your pictures look like?

Click to expand...

They look like this:

A bit over 1/200 of a second for the full sensor tracks with the maximum selectable continous shooting speed of 195fps. To me one of the major selling points of the camera is that sensor read speed to allow me to just freely use the eletronic shutter.

Horshack said:

A few months ago I started an open source / crowd source project on GitHub to create a definitive repository of sensor readout speed measurements, for both stills and video. The project uses the built-in LED on an inexpensive Arduino board as a standardized light source with custom software that toggles the LED at a known fixed frequency (500Hz). This assures all the measurements are calibrated and accurate to the same source.

I just published my results for the R3, which can be viewed at the link below. All of the columns in the live table are sortable by clicking on the column header - for example you can sort by photographic readout speed, 4K30p readout speed, etc... You can also toggle between milliseconds and 1/sec displays of the results. And also filter the results by entering a search pattern.

Canon R3 Sensor Readout Measurements

More detailed R3 measurements:

Canon R3 Sensor Readout Measurements (detail view)

For those interested here is the GitHub homepage of my project - it has lots of foundational information about rolling shutters, along with more details on the methodology employed for the measurements:

https://github.com/horshack-dpreview/RollingShutter

Click to expand...

1/208s on a 24Mp sensor .... a bit slow?

if that sensor were 45Mp it would be around 1/150s

compared to

Nikon Z8/9 1/268s with 45Mp

Sony A1 1/255s with 50Mp

Higgins2002 said:

Horshack said:

A few months ago I started an open source / crowd source project on GitHub to create a definitive repository of sensor readout speed measurements, for both stills and video. The project uses the built-in LED on an inexpensive Arduino board as a standardized light source with custom software that toggles the LED at a known fixed frequency (500Hz). This assures all the measurements are calibrated and accurate to the same source.

I just published my results for the R3, which can be viewed at the link below. All of the columns in the live table are sortable by clicking on the column header - for example you can sort by photographic readout speed, 4K30p readout speed, etc... You can also toggle between milliseconds and 1/sec displays of the results. And also filter the results by entering a search pattern.

Canon R3 Sensor Readout Measurements

More detailed R3 measurements:

Canon R3 Sensor Readout Measurements (detail view)

For those interested here is the GitHub homepage of my project - it has lots of foundational information about rolling shutters, along with more details on the methodology employed for the measurements:

https://github.com/horshack-dpreview/RollingShutter

Click to expand...

1/208s on a 24Mp sensor .... a bit slow?

if that sensor were 45Mp it would be around 1/150s

compared to

Nikon Z8/9 1/268s with 45Mp

Sony A1 1/255s with 50Mp

Click to expand...

I include a per-row readout time in my results, which normalizes the readout speed to resolution:

Higgins2002 said:

Horshack said:

A few months ago I started an open source / crowd source project on GitHub to create a definitive repository of sensor readout speed measurements, for both stills and video. The project uses the built-in LED on an inexpensive Arduino board as a standardized light source with custom software that toggles the LED at a known fixed frequency (500Hz). This assures all the measurements are calibrated and accurate to the same source.

I just published my results for the R3, which can be viewed at the link below. All of the columns in the live table are sortable by clicking on the column header - for example you can sort by photographic readout speed, 4K30p readout speed, etc... You can also toggle between milliseconds and 1/sec displays of the results. And also filter the results by entering a search pattern.

Canon R3 Sensor Readout Measurements

More detailed R3 measurements:

Canon R3 Sensor Readout Measurements (detail view)

For those interested here is the GitHub homepage of my project - it has lots of foundational information about rolling shutters, along with more details on the methodology employed for the measurements:

https://github.com/horshack-dpreview/RollingShutter

Click to expand...

1/208s on a 24Mp sensor .... a bit slow?

if that sensor were 45Mp it would be around 1/150s

compared to

Nikon Z8/9 1/268s with 45Mp

Sony A1 1/255s with 50Mp

Click to expand...

In practice I don't think it makes a difference, at least as a still shooter, since it's more than fast enough to effectively eliminate rolling shutter effect. Personally I prefer the uncompromised framerate with 14 bit RAW vs a faster readout but seemingly a compromise in hardware when it comes to processing the files with the Z8/Z9 that results in JPEG only above 20fps. Granted the files are much larger, but the Sony A1 doesn't seem to have the same compromise. I'd imagine that during development, the Benchmark for the R3 was the Sony A9/A9 II with its 1/160 readout at 24 megapixels, before the release of the A1 that was the figure to beat in regard to that spec. By the time the A1 was announced I imagine the R3 specs were pretty finalized.

Horshack said:

A few months ago I started an open source / crowd source project on GitHub to create a definitive repository of sensor readout speed measurements, for both stills and video. The project uses the built-in LED on an inexpensive Arduino board as a standardized light source with custom software that toggles the LED at a known fixed frequency (500Hz). This assures all the measurements are calibrated and accurate to the same source.

I just published my results for the R3, which can be viewed at the link below. All of the columns in the live table are sortable by clicking on the column header - for example you can sort by photographic readout speed, 4K30p readout speed, etc... You can also toggle between milliseconds and 1/sec displays of the results. And also filter the results by entering a search pattern.

Canon R3 Sensor Readout Measurements

More detailed R3 measurements:

Canon R3 Sensor Readout Measurements (detail view)

For those interested here is the GitHub homepage of my project - it has lots of foundational information about rolling shutters, along with more details on the methodology employed for the measurements:

https://github.com/horshack-dpreview/RollingShutter

Click to expand...

Excellent work, thank you!

Would also be interesting to see the measurements for the future Canon cameras (e.g. R1, R5II) once they've been released.

Very interesting. Not just the results, but the explanation as well.

In regards to results its easy to see the steady improvements in readout speed for newer cameras & the big jump made possible by stacked sensors.

As I own an R8, those results, in particular, interested me.

The jump in readout speed, between the R6 (same sensor as the 1DXiii?) and the R6ii / R8 is quite marked, which in itself is pretty impressive, especially given the price point of the R8, but not remotely close to what stacked sensors can achieve. I'm not sure how much more room for improvement there is before having to move to a stacked sensor. Are we pushing the boundaries of what non-stacked sensors can achieve? At what point will all new cameras move to stacked sensors?

And with the current readout speed of the best stacked sensors, how much advantage (in regards to distortion from rolling shutter) does a global shutter have? The elimination of banding would seem to me to be the bigger benefit & perhaps in video use as opposed to stills.

Anyway, thanks for all of the effort. It was very interesting.

Myles Baker said:

And with the current readout speed of the best stacked sensors, how much advantage (in regards to distortion from rolling shutter) does a global shutter have? The elimination of banding would seem to me to be the bigger benefit & perhaps in video use as opposed to stills.

Click to expand...

Relatively fast rolling shutters will always have some kind of lighting that still causes banding with higher shutter speeds; faster rolls just push the problem zone into faster strobing lights and faster shutter speeds. Someone in this forum just posted what looks like banding from a strobing light with the R3. The OP hasn't clarified everything yet, but it looks like a light in the scene may have been strobing at 66667 Hz.

The thread is titled "Did I reach the R3 sensor readout limits?".

A global shutter will not reveal strobing lights at all, in the form of banding, unless an object is moving fast through the frame. Just like you can see banding with your own eyes when you wave a thin white stick against a black background with a strobing light, a global shutter sees that stick with modulating illumination throughout its path, too. The camera, of course, is not responsible for it; it is just the reality of strobing light and a moving object.