Helium VS Alexa LF side-by-side test

[Laszlo Mohacsi]

I've been conducting a side-by-side test with my Helium and Alexa LF cameras. I did this test AFTER a shoot I did several weeks ago where I used the two beasts. That project came up all of a sudden, it required 2 cameras, so I didn't have time to make LUTs and I just had to go with it. I'll be posting that project soon, though for now, here are some examples of how the 2 cameras look when under the same shooting conditions:





Here's the setup for both cameras:

800 iso
24fps
5600K
180 degree shutter
T4

The Alexa LF was set to 4.5K Open Gate 2.39:1 ARRIRAW
The Helium was set to 8K WS redcode 8:1
The Helium OLPF was Low-light optimised (which I use 99% of the time).

Lenses were Leica-R (135mm on Alexa LF and 80mm on Helium) Naturally the field of view is different between the Full Frame Alexa and Super 35 Helium, so these lens options were somewhat of a match in terms of frame size

I used my Minolta spot meter to take the light reading at 18% grey on the colour chart, which was exactly T4

Here's what I found:

Using the false colour in the the Alexa LF (with LOG-C), the exposure was perfect at 18% grey. I usually don't use false colour on Reds because I rely on my spot meter, the histogram, goal posts and traffic lights. Everything looked good like I'd expect to see on the Helium's Red Touch. Though I did check false colour (with Log3G10) and the exposure was perfect at 18% grey also.

I also shot the same chart with both cameras set to 4K ProRes 422 HQ as a back up test... but RAW was the prime test

Just a quick note, I scaled down the Helium 8K to fit the frame of the Alexa LF 4.5K in the same timeline. Weirdly, the 2.39:1 Alexa (4448x1856) is a wider aspect ration than 8K WS Helium (8192x3456). 8K WS is 2.37:1

Here's a frame grab from the Alexa LF with ARRIRAW-to-LOG-C applied in Premiere, followed by a frame grab from the Helium with redcode raw to REDwidegamutRGB, Log3G10, output tone map = none, highlight roll-off = very soft

ARRI Alexa LF raw-to-log-c


RED Helium raw-to-log3G10


Straight up you can see the Helium is darker and with more "orange" in the reds. Take a look at the red camera tape on the back of the slate on the left side of frame. Other than this, there's very little differences.

On my 4K monitor and 16'' Macbook Pro, both cameras look beautiful, clean and easy to work with. The 16" Macbook Pro plays back both 8K and 4.5K raw video in the premiere timeline without lag.

Screenshot of raw settings:




I've been shooting with Reds and Alexas for 10+ years. My gut feeling has always been that Red 800iso is more like 500iso. I've never done a scientific test and never needed too. I've never had an issue with exposure on either camera system. I usually just "over expose" both cameras by 2/3 of a stop (just like I used to do on 35mm film), and get solid results. I rarely correctly expose to 18% grey and avoid white clipping through proper lighting.

Screenshots of lumetri color with scopes and settings in LOG-C and Log3G10:




Anyway, to get both the Helium and LF to match exposure, I applied the default Arri Logc2rec709 LUT to both cameras then boosted the iso in the Red source settings to 2000iso and added 6.8 points in the magenta tint (what's odd about this is the false colour tool in the Helium... it displayed green on the 18% grey colour chart, but is underexposed. If I set my spot meter to 500iso and used the iris setting it told me, which would be T2.8 1/3, I'd have a much closer exposure to the Alexa than I currently have). UPDATE: I've retested this theory and corrected my comments down below. 800iso on Helium is 800iso (not 500iso)

Here's the screenshots from lumetri color showing the scopes and settings. The red camera tape is still "orange" on the Helium (and yes, it was red to the eye and rather damn close to what the LF is showing).

Alexa LF with default Arri Logc2rec709 LUT


Red Helium with default Arri Log2rec709 LUT


The scopes show the Helium could still be boosted in exposure a touch more. The noise at 2000iso on the Helium is not noticeable (unless you have extremely poor lighting conditions), so this doesn't bother me. However, for best results, it's good to set the camera at 800iso, then meter at 500iso (or even 400iso) for a super healthy raw image. Here's the frame grabs with the Helium rescaled to 4.5K to match the frame of the LF

LF


Helium


My thoughts on this test:

- both cameras are excellent. In fact, they are my 2 favourite cameras on the market. I love the Monstro and Dragon too, but LF is brilliant for Full Frame because of the tall 1.44:1 sensor and Helium delivers 8K in super 35 (great for the Angenieux 12:1 and Panavision 11:1 zoom lenses, plus Zeiss Master Primes, Ultra Primes, Cooke S4s and Primos etc).

- I feel more than comfortable shooting with both cameras side by side in any lighting condition

- 8K is essential for 4K delivery, high end VFX, reframing and clean low-light performance.

- Red 800iso is closer to 500iso (i've now corrected this after more testing, 800iso is 800iso! See my comments below)

- Alexa LF is true 2.39:1 wide screen where as Helium is 2.37:1

- Alexa LF DOES NOT have a 4.5K 16:9 recording option (this is a frustrating problem in my opinion). The only way to get 4.5K 16:9 is to shoot Open Gate 1.44:1 then crop top and bottom.

- Helium performs well with the Alexa logc2rec709 LUT. If you're a colourist, feel free to grade the frame grabs I've posted and see what results you get. I supposed I could upload the raw video somewhere also.

[Mike P.]

Thanks for posting this Laszlo!

A couple questions; were you using ISOCal2 on the Helium? That's supposed to get "ISO800" to actually be a metered 800, whereas ISOCal1 is ~1 stop lower than that.

Conversely is there an option to "bypass IPP2 transforms" in Premiere (rather than using "none/very soft" for tonemapp/highlight roll-off)? Going with none/very soft is not the same as straight Log3G10, which may be why the metering/false colour were different on set compared to in post (and may make up for that having to boost to ISO2000 in post.)

Other than that (if you get the opportunity), it'd be great to see a Helium shot exposed at ISOCal2 1600 compared to an LF at 800 shot. That would give a closer even over/under on the Helium, which *may* make dragging and dropping LUTs/Looks between the two an even easier process (and it would also make up for the seemingly lost stop between the two if it isn't IPP2/ISOCal related).

OH, and can you post any raw frames and/or ~3second clips? (not like gigabytes worth, just quick/couple hundred meg files). I'd love to see how close they could get with Resolve's CST (see how it handles that tape orange/red).

[Laszlo Mohacsi]

Hi Mike, thanks for the questions. I'm usually working with a post house and a colourist etc due to the work I do, so my personal Premiere experience is limited. I have Resolve but wouldn't know where to begin with using it. So, I'll try to answer your questions as best I can (which are great by the way and have given me some extra insight into what I must do to nail the intercutting of Red and Arri cameras on my future projects). I'll do another round of tests in a few days which will include adding my Epic-X Dragon to the mix. I don't have a Komodo yet, but when I buy it, I will chuck it next to the LF and get a LUT/workflow made for seamless matching.

Thanks for posting this Laszlo!

A couple questions; were you using ISOCal2 on the Helium? That's supposed to get "ISO800" to actually be a metered 800, whereas ISOCal1 is ~1 stop lower than that.

Yes, I used the new ISO calibration:

Conversely is there an option to "bypass IPP2 transforms" in Premiere (rather than using "none/very soft" for tonemapp/highlight roll-off)? Going with none/very soft is not the same as straight Log3G10, which may be why the metering/false colour were different on set compared to in post (and may make up for that having to boost to ISO2000 in post.)

I'm not sure. A more specialised user of Premiere can probably help answering this, though as for now... this seems to be where the problem lies. I set the "Output transforms" to LOG3G10 etc for my test to achieve maximum dynamic range through the post-pipeline and to maintain LOG for as long as possible, before moving to REC709 for the final grade. This is my usual theory for working anyway, but correct me if I'm wrong. When I switch the "output transform" to BT.1886 the exposure at 800iso setting appears slightly brighter than the Alexa LF, though with more contrast (and therefor less dynamic range... right?). Anyway, to my eye, even if I use LOG3G10 at this point in the post-workflow, and boost the Helium to 2000iso, I can't see a noticeable increase in noise. In a 4.5K timeline, both the Helium at 2000iso and the Alexa LF at 800iso appear to have the same amount of noise (which is basically unnoticeable). Though I don't know how it would look in an 8K timeline because I don't have any solution to see a true 8K image.

Just for the sake of argument, I've shot several movies on Alexa LF and Red Dragon and never experienced an exposure or noise issue. Granted, we have budgets, access to adequate lights and post etc. I only bought my Helium a few months ago and have yet to shoot a movie with it. Though, I expect it to perform beautifully.

Other than that (if you get the opportunity), it'd be great to see a Helium shot exposed at ISOCal2 1600 compared to an LF at 800 shot. That would give a closer even over/under on the Helium, which *may* make dragging and dropping LUTs/Looks between the two an even easier process (and it would also make up for the seemingly lost stop between the two if it isn't IPP2/ISOCal related).

Good suggestion, I'll do this and a more extensive range of ISO tests on all cameras, though as mentioned above, it appears to be the "output transform" in premiere which is dragging the correct exposure down.

OH, and can you post any raw frames and/or ~3second clips? (not like gigabytes worth, just quick/couple hundred meg files). I'd love to see how close they could get with Resolve's CST (see how it handles that tape orange/red).

Yes I can, but I don't know where those files can be hosted? Any suggestions. When I get the time, I'll put some of these tests on youtube in 8K/4K. I have an anamorphic test I did with the Helium and LF side by side (full frame anamorphic on the LF) which is super interesting to see.

[Ignacio Aguilar]

Thanks for sharing this.

[Phil Holland]

Just to slide in here with a couple of quick answers, RED's Log3G10 compared to Arri's LogC are two different gamma log curves, which means your Log images will have different tonal placement. Additionally REDWideGamut and AlexaWideGamut are different color spaces with a fairly significantly different approach. If plugging footage into a RED IPP2 or Arri Rec.709 workflow a well made Color Space Transform will "convert" both the Color Space as well as Gamma Curve to better match to either of the target spaces.

In this example, it appears basically that you're bosting ISO to account for this difference and tweaking manually, which is not a bad way of working, just not as controlled or what colorists would typically do.

[Mike P.]

I'll do another round of tests in a few days which will include adding my Epic-X Dragon to the mix.

Helium, Dragon, LF! That'd be totally rad! I'm still rocking Epic-Dragon, so it'd be nice to see how far/close they are. Granted, DSMC1 is locked to ISOCal1, which means it'll be a stop under at every ISO setting (and it'd need ~ISO3200 to be closer to even over/under... so definitely can't use STH OLPF for that :)

Anamorphic stuff is always awesome to see, too! Not necessarily for comparison, but just to see how they look. LOL!

Yes I can, but I don't know where those files can be hosted? Any suggestions. When I get the time, I'll put some of these tests on youtube in 8K/4K. I have an anamorphic test I did with the Helium and LF side by side (full frame anamorphic on the LF) which is super interesting to see.

As for hosting, WeTransfer.com is free up to 2gb, and it'll stay up for a week before being deleted. It'd work in a pinch, and better than paying (presuming there's nowhere else to host.). OH, also if you have gmail Google Drive allows you to select which folders to be public/unlocked and you can share that link (though it's not as straightforward as wetransfer).

In this example, it appears basically that you're bosting ISO to account for this difference and tweaking manually, which is not a bad way of working, just not as controlled or what colorists would typically do.

Yeah, but he was comparing Helium on-set and Helium in-post as having an 18% grey delta (Helium to LF wasn't until later, though they too have a difference for reasons you mentioned). That's why I was thinking it was a "None/Very Soft" vs. "Bypass IPP2 Output transform" difference, as those two things are *not* the same (it's never been explained why selecting none/none for Output Transform isn't the same as "bypass", but it's not). In any case, it's not the end of the world because they were matched physically during the shoot (to the 18% mid-grey).

[Laszlo Mohacsi]

Thanks for the interesting information guys. I've spoken with one of my colourists here in Belgrade and he says they use BT.1886 instead of LOG3G10 when doing the grade in Resolve. I gave him the footage from Helium and LF and he thinks he has matched them rather easily. He'll send me the results and I'll upload...

Though, forgive me for being behind the 8-ball on the IPP2 workflow, because I don't usually delve into the details of how a colourist works on my footage, but what is the path for me to get the best results in Premiere? The reason I ask is, in this part of the world, Corona has significantly impacted production and I've been asked to colour my own work on the past few commercials. I'm under the pump with shoots, so don't have the time to learn Resolve. What does "output transform" mean in Premiere? Can it not be bypassed as Mike is suggesting? This is the detail that seems to be altering my exposure and making the image darker than Alexa. ARRI doesn't have this option in their raw settings and it just looks like a normal exposure in the timeline.

Here's my in camera settings:



Here's 2 options for the "output transform":

Log3G10


BT.1886


My gut is telling me it should be LOG3G10 and then throw a LUT on in Lumetri as an adjustment layer. Right?

If this is the road to get the best result, which LUT from RED do you recommend I use? On youtube I found a dude recommending the following:

RWG_Log3G10 to REC709_BT1886 with HIGH_CONTRAST and R_4_VerySoft size_33 v1.13.cube

Though, this youtuber doesn't state what his "output transform" settings are... so I'm getting confused and lost. Actually, the more I look into this "output tranform" business, there's little to zero information available on the web.

Anyone got answer or recommendation? I'd love to get my Helium (and Dragon) looking shit-hot in the Premiere timeline just like my Alexa. It will help satisfy clients who ask stupid questions :)

[Laszlo Mohacsi]

Okay this is interesting... I went ahead and used this LUT with my initial settings from the day of testing

RWG_Log3G10 to REC709_BT1886 with LOW_CONTRAST and R_4_VerySoft size_33 v1.13.cube

As a refresher, here are my settings for the Helium:

800iso
24fps
5600K
180 degree shutter
T4

Screenshots of my source settings in Premiere:





With the LUT applied as an adjustment layer in Lumetri, the result looks like this:



VS Alexa LF with the default Alexa logc2rec709 lut




Red Helium

Alexa LF


It's now looking close WITHOUT having to boost ISO at all. Of course with some fine tuning this could be an identical match between the 2 cameras, but as for basic LUTs and offline edits, this is satisfactory in my opinion. Let me know if you don't agree :) Have I done this correctly?

The next step is to load this LUT into my camera and use it for on set monitoring.

[Laszlo Mohacsi]

By the way... perhaps a more detailed discussion about this and all things Red/Cameras/Lighting/Cinematography could be done on "Clubhouse". If anyone is using it, I'm keen to get a room going and brainstorm ideas. It would most definitely be cool to get some other users thoughts before I shoot the following test on Tuesday: "Helium, Dragon, Alexa LF and Alexa Mini side by side".

[Šabović Adis]

Ehm, on that Spyder Checker book, what color are the stripes that are not green - red or orange? :)