Phil Holland Weapon Test

[Stuart Graham]

Yeah, I meant G's ...so I meant: RG2/3 & 4 in other words RC should be read RG Thanks Brett!

As this is aRED forum RC2/3 & 4 are all known to us all..The tests were great IMO, I can see what Phil shot means to him & I.

I used Jeff's WEAPON #68 this week on a commercial here in London & it was a breath of fresh air..changing from 25fps @5.5K to 100fps @4.5K & 125fps @3.5K was a 10 second task, no black shading required!
The extra highlight headroom is very apparent...I do feel though that this chip could do with RC5 being developed as I felt RC4 let highlights that were all there in detail were on fire on my Red PRO 7". 9" & the old 5" too, RAW check toggle showed all the detail..the highlights were 5.5 stops over 18%..impressive! & this was with the LLOLPF (& yeah no traffic lights)
As you all know or have read, it's a little smaller & lighter too, mini mags are still tough to get out when warm, miss the SIDE HANDLE A LOT!
30 second boot is the same as DRAGON but a lot quieter, we were doing 2 & 3 min takes with sound & I only heard the fan once in a 30/85 degree set but the sound dept. couldn't 1.2Tb's later we rapped.

Jeff Brown added ACTION PRODUCTS rear parts which gave us battery hot swap, 4 x HD SDI's & also a block battery XLR option too...roll on Off Hollywood's OMOD please!

Overall: A great improvement! as Jeff B wrote this morning: DNXHD 36 would be a welcome second recorded format alongside RAW & a SIDE HANDLE please.

The shoot was like most a little sensitive to R3D's being uploaded to the net, another reason Phil's test is just great!

 

[Ryan Hamblin]

Hey Phil, Thanks for the test. Was wonder if you happen to have a prores422 clip in 4K shot in red log-film? As well as the matching 6k r3d. I would love to play around with how far each can be pushed just to get an idea of what is best for certain situations.

 

[Tommaso Alvisi]

I'm with Ryan!

Please Phil try to upload a short ProRes clip if you can, with its mating R3D...

I'm really hoping for no downscaling artifacts and to find out if all the latitude is retained when using RedLogFilm with ProRes...

Thanks!!!!!

 

[Jeff McCoy]

Perhaps in the future some of us SoCal-based camera geeks can pull together resources and take another crack at it....

I volunteer time + equipment!

 

[Michael Lindsay]

Hey Phil

Modified Zeiss Otus lenses

!!!???

Spill?..What kind of modified?

Thanks M

PS sorry if you have already talked about it...

 

[Daniel Soderberg]

Totally agree. The color rendition on the STH olpf is much better than the other two. Not too thrilled with the new Standard one. Low light optimized is second best in color. And if you've got any small flaring sources, STH is the only acceptable option still.

Still love Skin-Tone Highlight OLPF! Great examples Phil.

 

[Phil Holland]

Needed a small breather from this thread.

Has Red said not to, or could you please post just a single R3D frame from each of the chart sequences?

I might provide .R3Ds to download here. Single frames are manageable. Not real keen on sharing my raw footage these days, but let me mull that over.

Hey Phil, Thanks for the test. Was wonder if you happen to have a prores422 clip in 4K shot in red log-film? As well as the matching 6k r3d. I would love to play around with how far each can be pushed just to get an idea of what is best for certain situations.

That I'll do. Might put up a 2K ProRes 4444 XQ clip too. Might be able to hammer that out by Monday.

Still love Skin-Tone Highlight OLPF! Great examples Phil.

Totally agree. The color rendition on the STH olpf is much better than the other two. Not too thrilled with the new Standard one. Low light optimized is second best in color. And if you've got any small flaring sources, STH is the only acceptable option still.

Thank you gentlemen. I'm a bit wary of discussing anything color related at this point, but I'm up late (2:45am PST apparently, damn) and tweaking while comparing a bazillion test images. Not exaggerating. This will likely interest STH OLPF fans. We have 3D LUT support on Weapon, inside REDCINE-X Pro, and just about everywhere these days. I've made about 44 LUTs over the course of this last week for all sorts of reasons, but one thing I've been looking closely at is the spectral response of the different OLPFs and how they play with different light sources. You can indeed match them just fine, but there are nuances in particular in the highlight roll-off that are a bit tricky to handle. One thing I began to experiment with is creating a "STH OLPF Compensation LUT". Basically something I can throw on top of DC2 with either RLF or RG4 that will make the Standard and Low Light Optimized OLPFs look a bit closer to the STH in terms of it's overall color response. Still playing with the matrix, but here's a work in progress shot:



I've tested this out on all sorts of footage and lighting conditions. I'm pretty happy with where the skin tones are. There's subtle lighting variations in this particular test scene in the background, but looking at her face and chart I'm pretty pleased. I have a bit more stress testing to do, but I can see myself using this LUT in camera when rolling with the LLO or STD (STD?!... hmmm STRD?). Likely could come in handy with baking ProRes dailies in camera as well for certain situations. Might makes these transforms for all the OLPFs as I do see some merit to bringing the STH closer to the STD and LLO at times.

Here's another set of grabs with the same stylized look applied on top of that:




Have some shooting and brief rental next week. Still looking for a slot of time to do a little A/B Weapon/Epic Dragon comparison.

 

[Christopher Probst]

How about Wed at Camtec, Phil?

 

[Phil Holland]

How about Wed at Camtec, Phil?

I'll make that work.

 

[Christoffer Glans]

Phil: I'm interested in the Standard OLPFs color performance, but it seems that it's quite "magentaish" compared to the other two, what's your take on this?
And, coming from the OLPF thread in the Red Discussion, is it possible to get any notion on how that distortion is on the Standard? Since my findings there I'm probably aiming for the Standard OLPF as my "main" one, but not if it performs worse than the STH in that regard. Do you have a wide open detailed R3D or crop to see perhaps?

 

[Bobby Farokhzad]

I completely understand the reluctance to share Raw files... any self respecting cinematographer/director/producer would probably feel the same way. However, since this is a scientific test meant to be shared by professionals, having a test like this and not providing some kind of a high quality file for evaluation is almost counter productive. Reading through this thread it's obvious that there is some confusion and that is because we have to evaluate color, contrast, noise, compression, and other elements of cinematography based on internet streaming and compression. I still consider Phil one of the greatest contributors to this community and always appreciate all his time and effort but without a Raw file everything is just a guess game.

 

[Akin A]

Maybe there is sensitive info in his metadata. I have had confidential notes in my metadata and forgot to remove it before going into another project and it's annoying that I couldn't remove it. Unless I just don't know how.

Edit:
I'm referring to the Project Slate info in camera. Forgetting to remove that slate info before the next project can be annoying if it's confidential info. As far as I know, that stays in R3D metadata.

 

[Eric Santiago]

I love what Phil brings to the table.

I dont have a WEAPON and probably never will.

I learn so much just from his posts.

I am probably in the middle here but whatever happened to show and tell without being challenged?

I am sure there are positive ways of asking for more info from Phil.

To me he has always been clear with his intentions.

Keep it up brother, you are a great resource to REDUSERs like me

 

[Maik Müller]

I love what Phil brings to the table.

I dont have a WEAPON and probably never will.

I learn so much just from his posts.

I am probably in the middle here but whatever happened to show and tell without being challenged?

I am sure there are positive ways of asking for more info from Phil.

To me he has always been clear with his intentions.

Keep it up brother, you are a great resource to REDUSERs like me

+11

Impressive and important work again and again.

And Phil, maybe think about selling those transform-LUTs once you are happy with them? Taking that idea a bit further, with all the LUT-packages one can get nowadays and then your knowledge and perfectionism in mind i actually would really like to see such a thing coming from you, optimized for RED and all. Just saying.

 

[Björn Benckert]

+11

Impressive and important work again and again.

And Phil, maybe think about selling those transform-LUTs once you are happy with them? Taking that idea a bit further, with all the LUT-packages one can get nowadays and then your knowledge and perfectionism in mind i actually would really like to see such a thing coming from you, optimized for RED and all. Just saying.

I think all weapons should be sold with a spider checker board and cube and red cineXpro should have a quick lay grid on checker, pick mid gray, pick highlight on chrome ball, click pitch back in the pit under the cube and then export perfect calbiration 3D lut to camera. If one low ress proxy frame could come across to an IOS device via the wifi then the calibration lut could easily be done on the phone... Imagine that, set camera to a choosen iso and close enough kelvin shoot redlog film and then tap on your phone to do the rest and then get a perfect as balanced lut for the scene right into the meta data folder for each clip...Must be possible and not that hard to make, no? Mikael probably allready have it in fool control 2.

 

[Zeb B]

I completely understand the reluctance to share Raw files... any self respecting cinematographer/director/producer would probably feel the same way. However, since this is a scientific test meant to be shared by professionals, having a test like this and not providing some kind of a high quality file for evaluation is almost counter productive. Reading through this thread it's obvious that there is some confusion and that is because we have to evaluate color, contrast, noise, compression, and other elements of cinematography based on internet streaming and compression. I still consider Phil one of the greatest contributors to this community and always appreciate all his time and effort but without a Raw file everything is just a guess game.

Agreed that Phil is a huge asset to RED and certainly the most knowledgeable and the best source of insider info there is.

What is fundamental with this test was to make a basic choice:

1. It's Science based and therefore up for brutal peer review and full transparency including raw files and test methodology.
2. It's an Creative project showing your craft and therefore should be respected as such.

Seems like this test was a hybrid of both therefore the confusion

 

[Anton Shavlik]

I'm wondering, if it's possible to match the images made with each OLPF as well as this, is there a reason that Red's color science calibrations for each OLPF are as different as they are? I understand that a 3D LUT is potentially destructive to the image, but a calibration might not be depending on how it's done?

 

[Matt Ryan]

I'm wondering, if it's possible to match the images made with each OLPF as well as this, is there a reason that Red's color science calibrations for each OLPF are as different as they are? I understand that a 3D LUT is potentially destructive to the image, but a calibration might not be depending on how it's done?

Agreed. If it is so easy to match color to the STH (from the Standard and LLO olpf) then why is the color science not tweaked to be identical on all of them across the board. I'm not an engineer or into coding, but if I can grade the LLO image to match almost identically to Skintone-Highlight olpf color wise, there should be a color science update that could match them out of camera. That would be ideal if we could pick up a camera with any olpf and still get identical color. That would be ideal... Looking at Phils example of matching above, I don't see why we couldn't have that since they look identical in his example (color wise).

 

[Phil Holland]

I completely understand the reluctance to share Raw files... any self respecting cinematographer/director/producer would probably feel the same way.

I'm personally at a professional crossroads with this particular subject and this particular test. One of my first paid shoots with Weapon was indeed a camera test that was similar to this one and included about a dozen more experiments. That I was paid to do, I own those files and the studio owns those files. In this case what I'm showing here is a test I funded to share my results because I said I was going to do that. The cost of the actress, crew, studio space, and licensing the audio track is on me. Having these REDCODE RAW files in effect allows for users to turn around and sell this content as well, which I feel victim to 2 years ago. Worse, some rather egregious examples were put online, re-graded, and called "somebody else's own".

My overall hopes with these images are to show things we need to see, what's possible out of the box, and what else can be done if worked on further. Standard stuff when it comes to working on motion picture material. Some of this information I don't mind sharing, some of it won't be shared as it jeopardizes my "job".

Agreed. If it is so easy to match color to the STH (from the Standard and LLO olpf) then why is the color science not tweaked to be identical on all of them across the board. I'm not an engineer or into coding, but if I can grade the LLO image to match almost identically to Skintone-Highlight olpf color wise, there should be a color science update that could match them out of camera. That would be ideal if we could pick up a camera with any olpf and still get identical color. That would be ideal... Looking at Phils example of matching above, I don't see why we couldn't have that since they look identical in his example (color wise).

To put simply, it's likely because it actually isn't that easy. What I'm finding while digging deeper is variance across certain light temperatures and some weirdness the light sources themselves provide. I think what RED and Graeme has done with the Color Science and OLPF Profiles is provide the most "universal" tweaking to provide a nice image under a large variety of conditions. In my opinion the STH OLPF seems to have the best color response across the board, it's also the most heavily coated/layered OLPF. And we're talking slight differences in reality when it comes down to it, but they are there.

I personally know I shoot around 5600K, 3200K, 2800K, and those few odd ball conditions most of the time. From my tests I see that perhaps the OLPFs can be tweaked to match further if aiming for a specific kelvin, but then individual lights (the lighting technology itself) can also provide a level of chaos. More testing needs to be done. It might come down to making compensation LUTs for each OLPF around approximate kelvin ranges, don't know yet, still digging deeper. I'll be sharing these LUTs this week with those who can take deeper look either way.

The matched results I've focused on so far have been under daylight and tungsten conditions and then I have tested those looks on a variety of other lighting conditions. Short answer is thatthere is a difference across different types of light sources and temperatures. So I think the current profiles/Color Science are doing a fine job in creating quality images under those varied conditions currently. Is something more can be done, RED will certainly do it. If it's not realistic or practical to change anything, I don't truly see a reason why it would need to outside of standard Color Science advancements. We'll see.

 

[Alain M]

Thanks Phil! Your work is always amazing for all of us to learn...