DreiFish

I mis-spoke -- I have the D-Pro (newer), not the older D-Max. I started testing it, but only got half-way because of some issues with my flash trigger (waiting on a replacement). Anyway, I can share a few details: 1. Flash duration at full power is 1/300s. So.. not fast, but adequate. 2. Color temperature unmodified is 6400k 3. Guide number (in air) is GN20. 4. Beam is nice, wide, without hotspots (as you would expect from a circular flash tube strobe). Here's a photo, and next two are the YS-D3 (dome diffuser, also GN20) and Marelux Apollo 3 (no diffuser, GN29)

1. Personal choice, but with more powerful video lights, you kinda have to turn them off in between shots or you'll run out of charge after one dive. It's.. a learned skill, but definitely slows you down and can lead to missing shots. This is where ergonomics like on the Keldans really do help -- as do lights that have a 'remote control' of some sort (again, Keldans!) and can both be turned on from the camera handle rather then reaching out to each one of them to turn them on. 2. Especially with more powerful lights (15k lumens+) there are many subjects that are bothered by them. One limitation in using continuous lights vs strobes. 3. Red lights are for focusing for still shots where the subject will be lit by strobes. You can't really use them for video.

Excellent resource, @Alex_Mustard. A critical examination of the images (pixel-peeping, I mean 😄) does seem to support the conclusion that (a) the foreground sharpness (or maybe it's micro-contrast?) is not on the same level as a native fisheye and (b) the zone of acceptable focus is narrower, rendering more blurred out backgrounds. For example, look at the surface water in this shot (upper left corner) in this shot - linked, not copied! (F11, 32mm) The fish school is in focus, but the background is very blurred. Granted, the slight zoom and 1/30 shutter speed might contribute to that. Or this shot (F13, 22mm, 1/60s). Or this shot. Another shot to showcase foreground sharpness and background (F13, 30mm, 1/200s). Compare this with a somewhat similar photo I've taken with the GH5 and Canon 8-15mm. Personally, I don't really see a image quality improvement you'd expect to see with a native fisheye on full frame. Or this shot from Alex With a couple taken with the Canon R5C, 8-15mm fisheye at 15mm, F8 What were your subjective observations on this topic, @Alex_Mustard?

In addition to the flash duration tests and recycling times for repeated shooting, I've done some test shots/measurements (in air) of the Marelux Apollo 3's power and beam coverage (in M and MTL, at the different power settings). I've measured (really, estimated) the guide number by taking test shots at 1.4m distance with the flash at different flash powers (from 12 - 1) and compared them with same set-up with the YS-D3. I really don't see how Marelux measured this as a GN44 strobe in air. It is basically about 1/3 stop less bright (in the center) than the Sea & Sea YS-D3. The reason for that is that it casts a much wider beam (more like a circular flash) than the Sea & Sea, so the light it outputs is spread over a wider area. Not a bad result.. but also not a real improvement. It does have better recycling times, allowing you to shoot at the same frame rates as the YS-D3 but producing 1 to 1 1/3 stop more light at each frame rate. Below is the beam coverage (in air) with and without the diffuser. To my eyes, the diffuser reduces a lot of light (2/3 stop) without doing much to improve the beam quality. As for the spread.. the triangular arrangement of 3 linear flash tubes makes it behave as something like a hybrid between a linear flash tube strobe and a circular flash tube strobe. Not as even as a circular flash tube, but better than linear flash tube strobes. Marelux Apollo 3 - No Diffuser / Diffuser (at full power) - ISO 100, F22, 1.4m distance Sea & Sea YS-D3 No Diffuser / Flat Diffuser / Dome Diffuser (at full power) - ISO 100, F22, 1.4m distance. Strobe is turned 90 degrees so the flash tube is vertical to better showcase the flash beam shape.

Looking forward to see some real images taken by a master with these strobes, @Alex_Mustard -- and of course, your thorough review if you end up making one. Your contributions to the community carry a lot of weight and I'm sure have helped countless photographers in improving their gear (and more importantly skills :D)

I recognize that this is a fairly close-focus kinda scene, but even so, I'd say that at F13, FCP shows much less depth of field in this situation than a traditional fisheye or fisheye with teleconverter would. Would you expect to get such a narrow slice of the image in acceptable focus shooting a 15mm fisheye at F13? Not my experience so far.. So it's not just that there's no improvement in image quality, you're getting more zoom range at the expense of both image quality and depth of field, which is a pretty big tradeoff.

Thank you Sergio -- this is helpful. In the full size images, it does appear that the first one the focus is somewhere on the diver or the gorgonians in the background not lit by the flash. It's really hard to say much about the sharpness from this image because the subject in focus is both unlit and far in the background, so normal water diffraction robs some of the expected sharpness. Second image, (parts of) the foreground coral are reasonably sharp. But definitely depth of field is limited, as even the front part of that coral (in front of the point of focus) is blurry. So maybe field of acceptable focus is ~10-20cm in depth? EXIF data shows this shot as F13, 1/100s at 29.5mm focal length. So this is basically at the widest end of the zoom range. If depth of field is so limited at such a wide field of view.. then yes, the FCP acts very differently than a traditional fisheye lens. It has a much narrower zone of acceptable focus. It's going to be basically impossible to keep an entire scene (foreground and background) in focus.

You can draw the conclusion that in HSS mode the flash will, at best, put out the same amount of light as it would have in the same time period in non-HSS mode. So, for instance, if you shoot it at 1/250s in normal and HSS mode, the HSS mode will at best produce equivalent elimination, but, most likely, actually less. With a perfect HSS implementation, if you go to a speed that's twice as fast as the flash duration, you'll get 1/2 the full illumination. But that's the best you can get. In practice, you'll probably get less than half because the zenon flash is less efficient firing a series of short bursts than one long burst during the same time period. (assumption, but it seems logical to me-- happy to be corrected. Do zenon flashes actually put out proportionally more of their light output in the initial portion of a flash than the second half of the flash duration? This is a questions of physics that I don't know the answer to). I'm not sure the example you gave can be used to conclude much of anything, especially since you've added the "similar" qualifier. F8 is 3 stops brighter than F22 1/2000s is 3 and 2/3 stops darker than 1/160s. So probably you actually got about 2/3 to one stop less flash light on your subject in HSS mode than you did shooting 1/160s F8, but also 2/3 stops less ambient light. It's a wash -- and you would expect the results to look different. But they're not 'better' -- you don't actually create any greater difference in comparative illumination between the foreground subject and the background. If HSS allowed you to perfectly capture the full output of the flash over a shorter duration, then shooting at F8, 1/250s should put the exact same amount of light on the foreground subject as shooting at F22, 1/2000s. And allow te exact same amount of background ambient light to fall on the sensor. It would be a wash from an illumination perspective. (Though, granted, you'd get better resolution because of the wider aperture and more ability to freeze motion because of the faster shutter speed. So perhaps you'd got a sharper image without diffraction effects and motion blur. But the illumination would stay the same.)

I've been shooting full frame and m4/3 systems for more than 10 years, with fisheye, rectilinear lenses, the WACP-1, WWL-1 and WWL-C (on full frame). None of these DOF issues show up there. (Yes, I'm familiar with Interceptor's investigation as to how water contact optics act to essentially project the image further from the sensor proportional to their coefficient of magnification, thus actually decreasing DOF) The problem with the DOF theory is that there isn't any particular part of Sergio's picture that is sharp -- not the foreground coral subject, not the background with the diver. And there's no noticeable difference in sharpness between foreground and background. It's uniformly unsharp. It looks like either the focus is entirely off (front-focused or back-focused) or the lens/FCP combination has optical quality issues.

Same measurement done on the YS-D3 led to the following estimates: GN 32 - 1/500s GN 22 - 1/600s GN 16 - 1/800s GN 11 - 1/1500s

I've now received my YS-D3 and did some initial testing. Guide Number: Measured at ~32 in air. Very similar to Backscatter's underwater test. Color Temperature: 6100k with no diffuser, 6050k with flat diffuser, and 6000k with dome diffuser. More or less same as Marelux Apollo 3s. Flash Sync Time: I estimate around 1/500s at full power, (GN32), 1/600s at GN 22; 1/800s at GN16; 1/1500 at GN11 Recycle Times: Can do 12 fps and 6 fps at GN 4 and 3 fps at GN 5.6 (according to strobe setting -- didn't measure actual exposure) Diffusers: Flat diffuser consumes .5 stops of light; dome diffuser consumes 1.4 stops of light.

And now for Marelux Apollo 3 flash durations at lower power, Photos taken at 1.4-1.5 meters from the wall. Full Power - Level 12 (Advertised GN 44 -- Estimated GN from measurement, 32) = ~1/350s Half Power - Level 11 (Advertised GN 32 -- Estimated GN from measurement, 22) = ~1/600s Level 10 (Advertised GN 27 -- Estimated GN from measurement, 19) = ~1/700s Level 12 in MTL mode (Advertised GN 22 -- Estimated GN from measurement, 19) = ~1/700s

I'm back home now and I've had a chance to test the Apollo 3s MTL mode to see if the claim is true that you can shoot at 10fps even at 'full power' (setting 12) - GN 22 according to Marelux's MTL power chart. I'm sorry to say.. it's not. At least not on the strobes I received. The R5C has only 3fps, 6fps, 12fps and 20fps (electronic shutter only - no flash) burst modes. So I can't exactly test 10fps mode, but.. In 12fps, I had to dial the power down to setting #6 - GN 8 in MTL mode to achieve 12fps. In 6fps, I had to dial the power down to setting #7 - GN 9.5 in MTL mode to achieve 6fps. In 3fps, I had to dial the power down to setting #10 - GN 16 in MTL mode to achieve 3fps. I'm using Nauticam's manual flash trigger for Canon, but before you conclude this may be a flash trigger limitation, keep in mind that I can get the flash to fire in 12FPS if you dial down the power far enough. It's a flash recycling time limitation. Also, despite Marelux's claims, there are (subtle) differences in the exposure between shots. Less than a stop difference, maybe closer to 1/2 stop or 1/3 stop, but certainly noticeable.

I gotta hand it to you Adventurer -- I was about to come back with a somewhat snarky reply that you should try it for yourself before concluding it can't be done 🙂 Very graceful, and definitely no hard feelings. We all read things too quickly sometimes. But of course I was going to bring proof that the test can be done, so here it is. A series of images taken with the Canon R5C (nominal x-sync speed of 1/250s) in full frame mode at ISO 100, F22 (w/ 15mm fisheye). Ambient light is minimized and flash exposure is from 1 Marelux Apollo 3 shot at full power in Manual (not MTL) mode. Marelux rates this as GN44 (but keep in mind I have the diffuser on and a 1/4 CTO gel, which should drop that to GN32 or even lower) Distance from flash to wall is about a 1.4 meters or 5 feet. 1/50s 1/60s 1/80s 1/100s 1/125s 1/160s 1/200s 1/250s Up until this point, the shutter doesn't block any of the light, and the exposure remains the same. 1/320s Here we're starting to see that the shutter is faster than the flash sink speed. But the illumination remains constant in the lit portions of the image. 1/400 Here we already see not only a larger portion of the frame blocked by the shutter, but also the illumination of the brightest portion has dropped. Here are the raw histograms for 1/320 and 1/400 side by side. So already we're faster than the flash duration. Based on this, I'd estimate the flash duration for the Marelux Apollo 3s to be somewhere around 1/350s at full power. The rest of the series shows what you'd expect -- increased portions of the frame blacked out by the shutter, and decreasing illumination in the portions of the frame still lit. 1/500 1/640 1/800 1/1000 1/1250 1/1600 1/2000 And here are the histograms to illustrate the drop-off in flash illumination at higher shutter speeds where exposure is shorter than the flash duration. You should be able to use this method to measure flash duration at least up to 1/2000s. I'll shortly test to see the flash duration also at lower powers for the Apollo 3s.

I do hope that it's a compression or focusing issue.. because otherwise, err.. nothing in those pictures is particularly sharp, certainly not what I would expect from a Z8.

1. Cameras with faster flash sync speeds are the ones that primarily benefit. Today, the only full-frame systems that fall within that scope are the Sony A9 III (1/16000s?) and Sony A1 (1/400s full frame, 1/500s in APS-C crop). 2. Higher sync speed is primarily useful for sunball shots where you want to knock down the ambient light exposure as much as possible using shutter speed rather than ISO & Aperture, so that the flash illumination in proportionally brighter. (If you stop down your aperture for this shots, this also reduces the effective flash illumination.) 3. Higher sync speed is not all that important for confined environments with limited ambient light such as caves or inside wrecks. There, even a strobe with 1/100s full output would be fine since you can shoot at lower shutter speeds and 'freeze' action. 4. Backscatter (in their video reviews) has published the flash duration for some recent strobes. I have added these numbers to the Strobe Comparison table pinned at the top of this forum: Ikelite DS230/232 (round tube flash): 1/107s at GN29; 1/200s at ~GN24; 1/320s at ~GN19 Backscatter HF-1 (2x Straight Tube Flash): 1/440s at GN40; 1/900s at GN34 Sea & Sea YS-D3 (2X Straight Tube Flash): 1/600s at GN32 Retra Pro Max (round tube flash): 1/600s at GN22 5. You can easily test flash duration at home without any specialized equipment. Simply take shots of a white wall at increasing shutter speeds while keeping ISO and Aperture the same (you can even do it above the flash sync speed of your camera) and compare exposure to determine at what point the exposure drops (i.e. the shutter is open too little time to capture all the light output). Once your shutter speed exceeds the flash sync speed of your camera, part of the image will be black (exposure blocked by the shutter), but you can compare the exposure in the part of the image that remains lit. 6. High Speed Sync DOES NOT get around flash duration limitations. It only gets you around the flash sync speed limitations of the camera. In HSS mode, the flash fires continuous short durations bursts, to try to illuminate the scene during the shorter period when the shutter is open. In HSS mode, you can think of the strobe acting very similar to a continuous light source like a torch. The shorter the shutter is open, the less strobe light you'll register. So a strobe with a 1/600s flash duration like the Retra, if shot in HSS mode at 1/1200s for example, will basically put out only 1/2 the amount of light that you would get at 1/600s. There's no free lunch here. This also means that HSS doesn't give you any advantage for sunball shots if trying to shoot them at shutterspeeds faster than the strobe's flash duration. 7. I have a Marelux Apollo 3, Inon S-220 and Supe D-Max I plan to test the flash duration for shortly. 8. Flash duration is roughly proportional to power output (in the same strobe). So if flash duration is 1/600s at full power, it should be around 1/1200s at 1/2 power, and so on.

I have the Marelux Apollo 3 strobe, but no way to replicate Backscatter's particular test without a similar chart I could put in my pool and some information on the distance from strobe to chart. What I will say is that the difference in coverage between the strobe without diffuser and with its diffuser is not really discernable on a real underwater scene. So probably the coverage is actually quite even even without the diffuser. Here are two pictures taken, one with diffuser, one without. Without Diffuser, full power, 1 meter distance, 15mm fisheye, ISO 100 F16 With Diffuser, full power, 1 meter distance, 15mm fisheye, ISO 100 F16

The flat diffuser, right, not the dome diffuser picture?

I think Backscatter should be commended for doing the test in water. Are we not perhaps over-hyping the reliability issue? While I agree that all these strobes could produce award-winning images, what has led you to the conclusion that there's meaningful reliability differences between them that should be the basis of someone's buying decision? With the exception of the YS-D2s, which (at least in the original versions) did have significant reliability issues (I owned 4.. 3 of them failed and the 4th was on its way to failure), I haven't heard any complaints about the quality of any strobe manufacturer. So.. what reliability data should we be using exactly to pick between these strobes? I'm betting that just like any of them could get you very good shots, any of them would also be reliable as well in the absence of any evidence to the contrary.

For me, the real improvement is the white balance at depth and (to some extent) the reduced noise.

Dome diffuser or the flat diffuser?

Nice comparison. It would be great if AOI (or others) offered an advanced housing for the Insta360 Ace Pro like they have for the GoPro now. https://www.backscatter.com/AOI-GoPro-HERO-9-10-11-12-Underwater-Housing-UH-GPX

Yes, it would be nice to have a test of repeat exposures in high speed shooting. Marelux claims that their MTL mode is specifically designed to reduce that issue. It's a relevant topic.. but.. mostly for a niche case -- high speed action, not general reef shots. I think your example shots illustrate that subtle differences in light intensity and falloff that can be seen on a test chart (even underwater) may not be all that relevant or noticeable when dealing with real underwater subjects. That is food for thought. Probably any of the 6 strobes I listed would be 'good enough' for most typical use cases.

Do you recall how close you were to the subjects? These both look to be somewhat close focus wide angle scenes, where the subject is centralized in the shot. It's exactly the type of shot where both narrow beam angle and 'hotspotting' would be hard to notice because there's no subject to be illuminated near the edges of the frame and the coral is all different colors and textures anyway, and at various distances, so it masks light unevenness. The test shots I posted earlier were at 1 meter from the subject, and the subject was basically a flat wall that more clearly shows variations in light quality. That's why I picked it -- not for any artistic merit. I'd also argue that what the exposure of the test shots is pretty spot on (at least for my taste -- some may prefer brighter subjects). Here's the histograms (Marelux first, S-220 second). The muted reds aren't an issue of exposure -- they're an issue of distance. The strobe light needs to travel through 2 meters of water to the subject and back, and that will filter out a lot of the reds at a proper exposure level.

I try to avoid taking such photos, but here's a comparison shot showing the 'hotspot' effect of undiffused Inon S-220s vs Marelux Apollo 3 strobes with diffuser.