HSS will minimize Backscatter

[Adventurer]

„HSS will minimize Backscatter compared to a single strong pulse of the underwater flash.“

 

What do you think about this?

[TimG]

Interesting thought. Where did this come from?

[humu9679]

Assuming all variables being the same - strobe angle, conditions, distance to subject, adjusted exposure - how can that be?

[Tobyone]

What is "HSS"?

[bvanant]

High speed sync could theoretically reduce backscatter depending on many things. HSS works by using a bunch of very short pulses. If the particles that cause the backscatter move between the HSS pulses then you could imagine that BS (that's what this is by the way) might be reduced. In practice in a BW dive in quite snotty conditions, not much reduction in backscatter.

Bill

[Tobyone]

If each flash froze the one backscatter particle , wouldn't multiple flashes create more points of reflection from the individual particle?

[Klaus]

Yes, but they are dimmer in absolute terms. What counts, though, is the relative intensity when the backscatter is compared with the subject. HSS compensates the fact that at faster exposure speeds, the shutter curtains move as an open stripe from top to bottom and multiple strobes (as in stroboscopic) illuminate the sensor as the curtains travel down.

Thus:

The backscatter-causing particles and the subject will receive the same amount (relative to each other) of light in both cases - HSS and « normal » strobe. If the backscattering particles move in an HSS relevant way, i.e. extremely fast  (we’re talking about shutter speed above 1/250 nowadays), then there is too much current and you should get out of the water asap. It should also blur the particles in a similar way for the standard strobe, if not more (full dump is longer than the stroboscopic action). And apart from solid structures, your subject will likely experience the same ripping current. It’s unlikely to be a great shot…
 

I think there are more important reasons to use HSS than backscatter reduction, but in the end it’s user experience than counts - and I cannot provide that for underwater. But top-side I can attest that HSS is not the solution for when it snows 😞

 

 

[Chris Ross]

Sounds like the claim is that each pulse of HSS light is very short and if the particle moves between pulses it will receive less light.  The problem is that HSS pulses are at very high frequency in the kHz range and likely the particle won't move very far between pulses. 

 

A quick analysis, assume a 20MP m43 sensor shooting at 0.3x .  Pixel pitch is 3.3 microns and the backscatter particle needs to move one full pixel which is 3.3/0.3 = 11 microns to not appear on the same pixel at the next pulse.  Assume the HSS is operating at 20 kHz so each flash is 1/20000 = 0.00005 sec.  To move 11 micron in this time particle needs to move at 0.000011m/0.00005 sec = 0.22 m/s.  Not particularly fast, that is 0.42 knots. 

 

However the particle is probably bigger than 11  micron to be conspicuous, lets say it's 1mm across so it needs to move 1/2 a mm so it doesn't cover any of the same pixels in the next pulse, so that's 0.0005mm/0.00005 sec, so that's 10 m/sec which is 20 knots.  Any slower and it leaves a short trail as it is illuminated each time the flash pulses. 

 

This analysis is not exact of course and only gives you an idea of how quickly things need to move to be not be recorded by more than one flash pulse.  In a HSS image at 1/250 shutter speed and HSS at 20 kHz, the total time the shutter is open is 0.004 sec and at 20 kHz that's 80 flashes.  You do get a moving window of the focal plane shutter and  that means each bit of image receives only some of these pulses, but the total light must be the same if the subject is to be exposed correctly.

[Klaus]

I think you need to factor in the optics in the calculation - 11 um on the sensor is more in the water, unless you are shooting 1:1 macro. I would assume that the need for HSS occurs more frequently at the fish portait-to-reefscape scale, so then the particles need to be more like 10x these speeds?
If this really made a difference, the forum would be full of examples and our GAS would be over and done with already… 

[Chris Ross]

33 minutes ago, Klaus said:

I think you need to factor in the optics in the calculation - 11 um on the sensor is more in the water, unless you are shooting 1:1 macro. I would assume that the need for HSS occurs more frequently at the fish portait-to-reefscape scale, so then the particles need to be more like 10x these speeds?
If this really made a difference, the forum would be full of examples and our GAS would be over and done with already… 

I've already factored that in by calculating the the amount a particle would need to move at 0.3x magnification to move one pixel width on the sensor which is 3.3 microns/0.3 x = 11 microns.

[Klaus]

But for a reef scene it would be 0.03 or even less - 0.3 is 1:3 so still a close-up, right?

Whatever, those who can shall try but I seriously doubt it’s worth the effort.

[humu9679]

16 hours ago, Chris Ross said:

However the particle is probably bigger than 11  micron to be conspicuous, lets say it's 1mm across so it needs to move 1/2 a mm so it doesn't cover any of the same pixels in the next pulse, so that's 0.0005mm/0.00005 sec, so that's 10 m/sec which is 20 knots.  

Diving conditions may be terrifying to bring this theory to practical use. 

[bvanant]

Actually I was thinking of Brownian motion, of relatively small but reflective particles. Certainly particles can move on the order of a few cm/s. If the particle moves enough rotationally then you might guess that it will have different reflectivity for different orientations. But remember I did say it was BS.😁

Bill

[Adventurer]

18 hours ago, bvanant said:

Brownian motion

I think everyone’s intuition that particles or subjects have to move to emit less backscatter reflection is not mandatory.

 

If you are familiar with astro photography image stacking you might develop a good intuition on what might be enough for the initial statement to be true.

 

In HSS the light pulses are usually less strong so they will cause a less strong reflection compared to a classic single strobe pulse (equaling the quantity of light that all HSS pulses together have).

 

As the shutter travels down (line by line)

during the (less strong) pulses, less backscatter reflection is captured, compared to a single strong strobe burst, giving out all at once and not in tiny slices.

 

[Chris Ross]

7 hours ago, Adventurer said:

I think everyone’s intuition that particles or subjects have to move to emit less backscatter reflection is not mandatory.

 

If you are familiar with astro photography image stacking you might develop a good intuition on what might be enough for the initial statement to be true.

 

In HSS the light pulses are usually less strong so they will cause a less strong reflection compared to a classic single strobe pulse (equaling the quantity of light that all HSS pulses together have).

 

As the shutter travels down (line by line)

during the (less strong) pulses, less backscatter reflection is captured, compared to a single strong strobe burst, giving out all at once and not in tiny slices.

 

This is true, but the same strobe light is illuminating both the subject and the backscatter, so if the backscatter doesn't move the pixels where it is recorded receive the same relative amount of light as the subject portion that is occupying the same slice of the image where the shutter is open.    If the subject is the same brightness in the final image the backscatter must also be the same brightness.  The only way it can be less bright is if it is recorded again on a different pixel or group of pixels on the sensor.

 

Another way to think of it is that the very high frequency pulsing on and off approximates continuous lighting so the result is similar to increasing the natural light coming in albeit from different directions and coverage.

 

To take the astro photo stacking analogy, if you get a satellite trail through one frame of a 20 frame stack it's only 1/20th as bright in that stack as it was only recorded on those pixels in that frame.  But you don't get that in backscatter unless the particle moves enough to not be recorded again on those pixels in the next frame.