Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Epic Dragon OLPF Test (STH vs LLO)

A

Akin A

Well-known member
Joined
May 3, 2014
Messages
868
Reaction score
0
Points
16
I tested STH and LLO, in prepping for some projects and wanted to share.

Overall conclusions:
- LLO appears about 2/3 (0.66) stop cleaner than STH.
- STH holds about 1/2 (0.5) stop more in highlights than LLO.
- STH has a green tint that needs adjusting.
- LLO has a warm look that needs adjusting.
- Both OLPFs can be matched to each other in post.

Experience:
- It takes me about 6 minutes to change OLPFs, including safely/carefully unmounting and securing the lens/lens mount/olpf and remounting everything back again, if I'm being extra careful with my stuff. I can do it faster though, at the risk of losing those screws or dropping something. It's a very easy process, just kinda annoying.

Worth Discussing:
- LLO is not a full stop cleaner according to my testing. I've read anywhere from 1 to 1.25 stops, but I haven't been able to replicate those results.
- It's been said that STH "blocks" more light than LLO and that LLO is "more sensitive" to light than STH. The exposure levels are the same for both OLPFs, with the same camera settings and lighting settings. Does this mean that the camera is remapping middle gray, based on which OLPF you choose in the settings, as suggested by Anton in a reply below?

If I'm wrong about anything or tested improperly, let me know. Just trying to learn more and share the knowledge.
---

Test #1A
OLPF ISO Test (STH vs LLO) REDCODE 4:1
https://youtu.be/yePbj4tUeMU

This is an ISO test using the Skin-Tone Highlight OLPF and the Low Light Optimized OLPF on the RED Epic Dragon.

YouTube's compression is making the footage look way cleaner than it actually is.

My opinion of the noise levels, based on the actual DNxHD export from Resolve:

Skin-Tone Highlight:
200 - 320: clean
400 - 640: 6K crop: looks like fine grain, 2K downscale: clean
800-1000: 6K crop: noticeable, 2K downscale: clean/fine grain
1280 - 6K crop: slightly distracting, 2K downscale: fine grain
1600 - 6K crop: distracting, 2K downscale: fine grain
2000 - 6K crop: distracting, but usable with denoising, 2K downscale: fine grain
2500 - 6K crop: very distracting, maybe usable with denoising, 2K downscale: noticeable
3200 - 6K crop: no, 2K downscale: slightly distracting
4000 - 6K crop: hell no, 2K downscale, distracting but usable with denoising

Low Light Optimized:
cleaner by 2/3 stop at each ISO

Lens:
Rokinon 85mm F/1.4
Apertures:
F/1.4 (ISO 250 - 640)
F/2 (ISO 800 - 1,000)
F/2.4 (ISO 1,280)
F/2.8 (ISO 1,600)
F/3.4 (ISO 2,000 - 2,500)
F/4 (ISO 3,200)
F/4.8 (ISO 4,000 - 5,000)
F/8 (12,800)

Resolution: 6K FF 6144x3160
- downscaled to a 1080p timeline.
REDCODE: 4:1
Shutter: 1/48s
Color Temperature: 5600K
Color Space: DRAGONcolor2
Gamma Curve: REDgamma4

Light:
Key: 1 Westscott Flex LED 10"x10" 55W 5600K at various intensities to expose for each ISO

NLE: Davinci Resolve 12
- Export: DNxHD 1080p 220/185/175 10-bit
---

Test #1B
OLPF ISO Test (STH vs LLO) REDCODE 8:1
https://youtu.be/cEE2drprzT8
I originally shot this one first and didn't realize I left my camera on 8:1 compression. It's still a good test, so I'll leave it.
4:1 compression is a little cleaner than 8:1. You might not notice through YouTube.

My opinion of the noise levels, are the same as test #1A, above.

Lens:
Rokinon 85mm F/1.4
Apertures:
F/2.8 (ISO 250 - 2,000)
F/4 (ISO 2,500 - 4,000)
F/5.6 (ISO 5,000 - 6,400)
F.8 (ISO 12,800)

Resolution: 6K FF 6144x3160
- downscaled to a 1080p timeline.
REDCODE: 8:1
Shutter: 1/48s
Color Temperature: 5600K
Color Space: DRAGONcolor2
Gamma Curve: REDgamma4

Light:
Key: 1 Westscott Flex LED 10"x10" 55W 5600K at various intensities to expose for each ISO
Some changing sunlight coming from window lit the back wall, behind the monitor

NLE: Davinci Resolve 12
- Export: DNxHD 1080p 220/185/175 10-bit
---

Test #2
OLPF Highlight Roll-Off Test (STH vs LLO)
https://youtu.be/5S7IiYCgxuY

This is a highlight roll-off test using the Skin-Tone Highlight OLPF and the Low Light Optimized OLPF on the RED Epic Dragon.

I brought the exposure back down 6 stops using the FLUT adjustment, in order to better show the actual clipping. This is not meant to look pretty.

Lens:
Rokinon 85mm F/1.4
Apertures: F/2.4, F/2.8, F/3.4

Resolution: 6K FF 6144x3160
- downscaled to a 1080p timeline.
REDCODE: 8:1
Shutter: 1/48s
ISO: 800
FLUT: -6.0 (to better show the clipping)
Color Temperature: 5600K
Color Space: DRAGONcolor2
Gamma Curve: REDgamma4

Light:
Key: 2x Westscott Flex LED 10"x10" 55W 5600K 95 CRI at full intensity

NLE: Davinci Resolve 12
- Export: DNxHD 1080p 220/185/175 10-bit
---

Test #3
OLPF Color Match Test (STH vs LLO)
https://youtu.be/0VY-oYSeTfY

This is color matching test using the Skin-Tone Highlight OLPF and the Low Light Optimized OLPF on the RED Epic Dragon.

The STH OLPF has a green tint. The LLO OLPF has a warm look. I manually adjusted the white balance and tint by eye to get the two to match. I only wanted to use the RAW adjustments to get a basic match, but I could have used primaries and secondaries to be more accurate. YouTube is making the colors look more cool (blue) than it does in Resolve.

I didn't use Resolve's Color Match feature on the X-Rite Colorchecker Passport because at the moment, it doesn't look good with RED footage.

Lens:
Sigma Art 50mm F/1.4

Resolution: 6K HD 5568x3132
- downscaled to a 1080p timeline.
REDCODE: 8:1
Shutter: 1/48s
ISO: 800
FLUT: -1
Color Temperature: 5600K
Color Space: DRAGONcolor2
Gamma Curve: REDgamma4

Light:
Key: 2x Westscott Flex LED 10"x10" 55W 5600K 95 CRI

NLE: Davinci Resolve 12
- Export: DNxHD 1080p 220/185/175 10-bit
 
Last edited:
You're welcome, Nick.

I added a 4:1 compression test to the original post and updated some results.
 
I think the sensor, in the end, is the same, and that's why there are people saying that the LLO lets more light through. And they're right.

The ISO is adjusted based on that, which it should be since the S in ISO is 'standard'. If one sensor+OLPF combo is half as sensitive to light, we should rate it at half the ISO, and that's exactly what it seems that red did.

In other words, with the LLO installed your .r3d is mapping middle grey differently than the STH for the same ISO. You would also expect that the noise levels jump up the same amount as the clipping point, but that might not be true because there are 3 channels, and the OLPFs seem to balance those 3 channels differently.

edit: also the STH demonstrably blocks more IR, which is very important for good colors. I've also noticed it adds a green tint across all the lenses I've used.

Of course I don't really know what's going on under the hood, that's just what seems to make sense to me.
 
Last edited:
Would be nice if Red developed a new OLPF, some kind of improved midstep between STH and LLO. Would be perfect. I love the STH but it can be a bit hard sometimes to get a clean image when I need to.
 
Christoffer Glans said:
Would be nice if Red developed a new OLPF, some kind of improved midstep between STH and LLO. Would be perfect. I love the STH but it can be a bit hard sometimes to get a clean image when I need to.
Click to expand...

Give me 2 minutes.
 
You must log in or register to reply here.
Back
Top