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 :)

Dragon has more DR than film?????

Brian Boyer said:
Thanks, Evin. That's good to know. I thought that should be the case but I haven't really seen a good example of it yet. To be fair, I don't think anyone has shot with that in mind yet.

Like Emmanuel stated, perhaps DRAGONCOLOR will improve upon it even further.
Click to expand...

Did you watch Peter's test? He shot the Santa Monica pier at night that has colored "christmas" lights (right side of frame) and they look great. No colorless center with colored halo. Check it out again if you are curious. Although they are quite a distance away... 00:59 second mark.
http://vimeo.com/72861708
 
Matt Ryan said:
Did you watch Peter's test? He shot the Santa Monica pier at night that has colored "christmas" lights (right side of frame) and they look great. No colorless center with colored halo. Check it out again if you are curious. Although they are quite a distance away... 00:59 second mark.
Click to expand...

Thanks Matt. I actually pointed those Christmas lights out in Peter's original thread:
http://www.reduser.net/forum/showth...Test-for-ASC&p=1246834&viewfull=1#post1246834

However, the traffic and taillights aren't faring too much better than MX in the night street scene of Gunleik's "Sweet..."

Like I said, that scene probably wasn't shot with those things in mind and maybe the new color science would better leverage the sensor's improved capabilities.
 
Actually, the traffc lights do even better than what the vimeo looks like.

I graded for DCP, and that did not translate too well into the quicktime. I had the choice of making "tonemapped", dull or let some of the highlights and lowlights go.

Now, a week later, i have a lot better tools for grading... Hahahaha.

Isn't it fun on the bleeding edge?

Looks good in the DCP, looks ok in h.264...

Right now i do not have time for a regrade, though...
"Should" do it. :-)
Must just pay the rent...

Best.

G
 
Gunleik Groven said:
http://www.reduser.net/forum/showthread.php?105573-Sweet&p=1260689&viewfull=1#post1260689
Click to expand...

Gunleik, thanks for posting these. You are a gentleman and a scholar. Buuuuut...

Unless I'm missing something, the linear light version isn't showing much improvement (color saturation-wise) in the taillights over the RG3 or REDlogfilm versions.

The car closest to camera shows the most color saturation, which is promising (yet nowhere near the deep reds you'd see on film), but the two cars in front of it are only showing a faint hint of color in their lights and are further away, which makes me think they have the brakes applied where the closest car does not.

That, or the closest car is at an oblique enough angle to help retain color saturation. A good example of this effect occurs at 3:24 in the video below. As the car turns the corner, the taillights become increasingly white as the angle becomes more perpendicular to camera.

I was hoping to see something from Dragon at least along the lines of the color saturation seen in the "Sca_dic" sign on the right side of the image in the linear light version.

It may be too much to ask of digital at this point but film reproduces brake and taillights with the saturation of the reflected red emanating from the inside of the traffic lights in the RG3 and REDlogfilm versions or the "TGIFRiDAYS" sign in the linear light version.

Additionally, the green of the traffic signal is still more of a white disc than a green blob - like the one further down the street, just right of center, but that may just be due to it being out of focus enough to appear green.

HDRx isn't really a help in this situation because the clipped white where the color should be doesn't allow for a satisfactory blend.

Perhaps purposely exposing so the taillights hold some color and taking advantage of Dragon's reduced noise floor to lift the darks would make the difference.

One thing that did surprise me was the amount of color that exists in the sign above the FRiDAYS sign that lists the 17 or so logos (store names? brand names?) seen in the linear light version.

http://vimeo.com/74549017

 
Last edited:
Thanks.

Actually the saturation is not best displayed in the linear, but clip is...

As to traffic lights, there is green in green and red in red. The ones appearing white are for trams, and are actually white... :)

I agree with your observations, I might make one more development of this shot just for fun, I have a hunch...

There is a bit that happens in development (even though I did not do anything but what I describe), and I am curious, too if there is more color there.

That things clip, does not worry me too much, really.

With a lower exposure, I would naturally hold more of the highlights... :)

I will check it later. I am not with the material atm.

Best

G
 
Gunleik Groven said:
As to traffic lights, there is green in green and red in red. The ones appearing white are for trams, and are actually white... :)
Click to expand...

I see the white of the white lights (arrows and uppermost circular lights) but the reddest and greenest parts of the red and green lights in this pic (and the shots from the video) are the reflections of the lights off the hoods covering them, not the lights themselves, which have the faintest tint of red or green.

I was hoping that with Dragon, the green of the directly viewed green traffic light would be as green as its reflection in the window by the FRiDAYS sign, above the trashcan, and the direct red lights (not seen in the still, but visible in the video) as red as their reflection in the window between the people walking.


I don't know the first thing about color science but it seems like it's tuned to work in >95% of shooting circumstances where you're capturing reflected light from the objects in your scene, whose highlights would always naturally desaturate and go white.

If that's the case, colored light sources present a unique problem for that tuning bias because their color would tend to saturate at equivalently bright reflected light levels instead of going white.

In addition, the colored light sources that go white aren't always clipping the sensor. It's like they're clipping the color science algorithm, if that makes sense.

Film has the photochemical thing going for it but digital may just have to rely on the brute force of wider dynamic range to overcome the problem unless there's some mathematical trickery that can be accomplished.

Other than through relative intensity, the camera would have no way of knowing if the red is from a traffic light or from a flower in bright sunlight.

Maybe a alternate version of HDRx could be implemented whereby the X-track is used solely for analytical purposes to determine the color and relative brightness of and between pixels compared to the adjacent A-Frames.

In RCX, a process that identifies a significant enough change in pixel brightness and color could be tweaked to replace whitish blobs with the appropriate colored blobs. It won't work in every situation but neither does the current incarnation of HDRx.

As stated before, Dragon's new color science may retain more color info than RC3 allows for because RC3 was tuned for MX, which has less DR and fewer bits.


I know it sounds like more trouble than it's worth but so many people get new cameras and immediately want to post video showing how well it performs in low light by shooting nighttime city footage.

They're so impressed by how much they can see into the shadows they ignore what, in my opinion, makes that environment so appealing to shoot - the interplay and richness of all the colors, which come from the various light sources.

Who cares if you can see the curb that falls in shadow underneath a car? Has anyone ever looked out over a cityscape at night and thought, "Look at all the dark areas."?

No, you look out and think, "These lights look so cool." Then, you grab your digital camera to capture that moment and get a weak representation of what compelled you to take the picture in the first place - white blobs with colored halos. It'll do, but I look forward to the day when it's no longer an issue.
 
Last edited:
It's known film, especially modern film, has really sweet overexposure latitude and it shows in cases like traffic lights, though to be fair I've seen somewhat similar "white blob and halo" effect in stuff shot on film, but graded without highlight protection in mind. Just saying.
 
Vadim Bobkovsky said:
It's known film, especially modern film, has really sweet overexposure latitude and it shows in cases like traffic lights, though to be fair I've seen somewhat similar "white blob and halo" effect in stuff shot on film, but graded without highlight protection in mind. Just saying.
Click to expand...

Yeah, I've seen it on film as well but not nearly as often. On the show "Glee", which is shot on film, the gelled lights often go white due to their extreme brightness during the performances they shoot on auditorium stages.

It happens, just to a much lesser extent.
 
Speaking of DR, can someone explain to me how S/N in decibels relates to dynamic range and how RED goes about calculating it?

I keep reading 6db = 1 f-stop but that doesn't match up to the DR RED publishes for their sensors. Based on this, Dragon's S/N should be 96db, not 80db.

What numbers are other manufactures using? When I read a sensor has 72db, I'm not sure how I should be calculating that into f-stops.
 
Brian Boyer said:
Speaking of DR, can someone explain to me how S/N in decibels relates to dynamic range and how RED goes about calculating it?

I keep reading 6db = 1 f-stop but that doesn't match up to the DR RED publishes for their sensors. Based on this, Dragon's S/N should be 96db, not 80db.

What numbers are other manufactures using? When I read a sensor has 72db, I'm not sure how I should be calculating that into f-stops.
Click to expand...

As a Sound Engineer I can tell you what happens with sound, which is kinda similar.

The thing is that the measurement is logarithmic.

For instance, if you go from 0dB to 30dB, no much happens, is similar to whispering in a quiet room.

But, from 30dB to 60dB the sound will be a thousand times louder, similar to be in a crowd with many people talking to each other.

From 60db to 90dB the sound will be a million times louder, which is similar to be in a Rock concert.

Now, 120dB, they call it "The Pain Threshold"; is when your inner ear's membranes break, and they say, is similar to be close behind the exhausts of an F32 raptor, or the Concorde with full power, and that's only 30dB more..

Now, how many dB's 3 stops represents for instance?, I don't know, you do the math, but I can tell you right now that, there are rumors that THIS sensor, its been used by Arri and Blackmagic on their cameras, which is 60dB S/N, and 90dB S/N with HDR, and is well known that they claim that their cameras have 12 stops of DR.
 
Last edited:
Strange rumor: neither Arri nor BM claim 4K photocells in current cameras.
 
Hii Rohan,
Very Soon we should have our First Dragon in India. Wait a little more & you'll see its power.
I had a opportunity to seethrough its LCD, one thing i can tell you it has got some reach & hold in its Histo.


Cheers...
 
You must log in or register to reply here.
Back
Top