Yet another Dragon SxS Helium Example

[Gavin Greenwalt]

Do. Not. Compare. Resolution. They are different F-stops. This is about dynamic range and noise. And even then take it with a thousand grains of salt. I just turned the Epic-W on, let it warm up for a half minute to green T for the very first time and then hit roll for 7 seconds. I haven't spent any time with Epic-W yet. Very much "first results".

Ok, so the first point that needs to be made is Dragon and Helium are quite different sensitivities so just going A/B is nearly impossible without a lot of Caveats however I got within about a quarter of a stop on the upper end of clipping and burned out two EXRs. The next giant caveat is that the color science will have a big effect on how the noise is digitally gained. I started trying to fudge the two to be nearly the same on color but then decided that it would just add yet another variable. Another wild card is that this is an Epic-Dragon not a Weapon-Dragon so RED has presumably also improved Weapon's Dragon already. Colorspace for this is from the REDWideGamut Colorspace, do what you will with that information. One last caveat is that one camera is on Skin-Tone (Dragon) and one is on Normal (Helium). I don't have a "normal" Dragon OLPF so more grains of salt to the wind.

Both were with the Canon 24-70 II @ ~50mm. Both were at 1/48th of a second. Both were at the 5:1 minimum compression of Epic-W. Dragon was handicapped on exposure by about 1/4th of a stop by my guess since I didn't hit the exact clip point on both cameras by eye on the viewfinder while iris ramping. I also scaled up Dragon to 8k before cropping/outputting to keep things a little easier to compare.

My take away is that Helium is without question more sensitive. And probably comparable to Dragon for dynamic range. I would be inclined in my absurdly limited testing to say Helium has a smidge more but it's going to be a very subjective "usable" argument one way or the other. Helium probably has at least 3/4 of a stop. Stacey Spears and I will be doing some less subjective real dynamic range testing soon...


TL;DR; caveats:

- Dragon is Epic Dragon not Weapon Dragon.
- Dragon is ST-OLPF, Helium is Normal-OLPF. Makes comparing sensitivity and color rendition a fool's errand.
- Dragon is underexposed by 1/8th of a stop compared to helium.
- F-Stop is F5 for Helium, F6.3 for Dragon.
- Focus is not exactly consistent between the two.
- Both were exposed to the same white clip point not for middle gray.

If you click on an image it'll take you to OneDrive and you can select "View original" to see the uncompressed png.

R3Ds: https://1drv.ms/f/s!Aq6vwny1LaQQqrQ9Rgg-dj8JxuraEg

Full Dragon PNG: https://1drv.ms/i/s!Aq6vwny1LaQQqrNpbP9qXrDxIMN9sg
Full Helium PNG: https://1drv.ms/i/s!Aq6vwny1LaQQqrNqdbSv2AAHk6UrVA

Dragon:

Helium:


Dragon:

Helium:

[Alex Boothby]

Thanks Gavin! Much appreciated. DO you know what the difference in F-stop was? Approximately? I started comparing noise, but then realized that's not what the test was about :) Presumably Helium would be cleaner / brighter if the stops matched.

[Gavin Greenwalt]

Thanks Gavin! Much appreciated. DO you know what the difference in F-stop was? Approximately? I started comparing noise, but then realized that's not what the test was about :) Presumably Helium would be cleaner / brighter if the stops matched.

No, noise is the one thing you can compare... kind of. :D RED didn't record the metadata per-frame for F-Stop so I don't know what the corresponding F-Stops are.

I exposed to match the clipping white point in both frames (but Dragon at about an 1/4th stop disadvantage) so you can compare the noise level to see which has more usable stops (easily helium but by how much?).

Ultimately this doesn't give you an exact A/B comparison but it at least shows that the two are definitely in the same ballpark. Helium looks a good bit cleaner even with a healthy margin of error in the testing. You can say "at worst, Helium has similar dynamic range to Dragon but more resolution, at best it's got cleaner shadows, more resolution and better sensitivity". I'll try to put some more exact numbers to all of those generalizations soon. With different OLPFs the sensitivity question is impossible to pin down exactly since the two OLPFs have different densities.

[Alex Boothby]

No, noise is the one thing you can compare... kind of.

Oh... yeah true... if your tendency is to expose for highlights. I'm so concerned about noise floor that I tend to expose for that. Helium will be cleaner than Dragon if you rate at 800 ASA, but it may be wise to rate at 1600 to hold those highlights.

You can say "at worst, Helium has similar dynamic range to Dragon but more resolution, at best it's got cleaner shadows, more resolution and better sensitivity". .

Well said.

[Hrvoje Simic]

No, noise is the one thing you can compare... kind of. :D RED didn't record the metadata per-frame for F-Stop so I don't know what the corresponding F-Stops are.


I exposed to match the clipping white point in both frames (but Dragon at about an 1/8th stop disadvantage) so you can compare the noise level to see which has more usable stops (easily helium but by how much?).

Thanks for sharing the results and for "side by side" instead od "vs".

Okay, so no r3d and no info on the stops used on both. Any info on exposure and ISO difference ? :)

[Michael Lindsay]

Dragon weapon or epic dragon?

Thanks

[Gavin Greenwalt]

Yeah, sorry about the no info on stops. I really just turned my camera on to make sure it worked. I kind of was hoping the per-frame Aperture info would come through now but I guess that's still not implemented into the R3Ds. And since they're different OLPFs a precise measurement would be pretty useless.

As to "ISO", I have honestly not the foggiest. I exported as EXR, adjusted the white point to ensure that both images had their clip point as 100% white, adjusted the black to the Dragon's black clip point and threw on a custom gamma. It doesn't actually reflect any "standard" RED ISO processing since it was a one-off image pipeline. I'm uploading R3D stills. You can play with them. Easiest place to confirm the clip points are similar is the width of the clip region on the edge of the table.

R3Ds: https://1drv.ms/f/s!Aq6vwny1LaQQqrQ9Rgg-dj8JxuraEg

Dragon weapon or epic dragon?

Epic Dragon and Epic-W

[Dawson L. Dunning]

thanks for posting these! just fyi, r3d says f5 on dragon, f6.3 on helium both ISO 800, helium is the one just under highlight clip point not dragon.

[Phil Holland]

Couple of odd questions.

Why 8K 2:1 against 6K FF? You've got a few more pixels to explore!

These aren't shot with the same OLPF. Any reason for that?

Just curious.

[Filip Orlandic]

These aren't shot with the same OLPF. Any reason for that?

One last caveat is that one camera is on Skin-Tone (Dragon) and one is on Normal (Helium). I don't have a "normal" Dragon OLPF so more grains of salt to the wind.

Answer :)