Welcome to our community

Be a part of something great, join today!

  • 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 :)

HDR From RAW FILE in RedCine

Joined
May 2, 2007
Messages
1,288
Reaction score
0
Points
0
Age
42
Location
Tbilisi , Georgia
Website
sarke.ge
Today i check with RedCine to make HDR . It can be funny bat its works. I now that it is not real HDR but but .
RAW file is fine to do small HDR effect. Also we check with moving of camera .
With help of Photomatix software , we just renew over old test footage . And it is nice in action .

Here is my First HDR pic from R3D file . Also i want to test with motion control system . We want to shoot motion 3 times and change only ND filters.
Sow it will be really HDR. Dose some body test it , Also i have no
good footage with sky to make HDRs , i will shoot it on this days.
Please don't forget that it is my first HDR in my live :)
Working With 8 Bit 1920 -1080 Tiff files.

Thanks.
 
On your way...

On your way...

Here is my First HDR pic from R3D file . Also i want to test with motion control system . We want to shoot motion 3 times and change only ND filters.
Sow it will be really HDR. ... i will shoot it on this days.

Sounds like you have a plan, please post the two or three source images that were fused in addition to the fused HDR result so we can see the exposure in each of the source images and how they mix in the result image frame.

You may need a IR filter with the ND so you do not get a color shift in the highlight exposures.

Good luck, look forward to seeing your results!
 
IR...

IR...

Can you give me more information about IR filter effect in HDR ?
Thanks .

There were some other posts showing I think what looked like a pink cast when some ND or Pola type filters were used, since you are only using ND0.3 and ND0.6 it might not be so bad. Up near ND1.2 or over some filters might pass enough IR to matter, it depends on the lights used. For static tests or stop motion you can change the shutter speed to get an ND effect without the extra IR.

Plastic ND filters, gels, or ones that use dye layer may pass more IR than mirrored ND filters, or IR coated ND filters. You can ignore the IR if it is not a problem in your test shots, I thought you might be going 5 or more stops down...

See also,

http://www.reduser.net/forum/showthread.php?t=9412

And about filters to fix the issue,

http://www.reduser.net/forum/showthread.php?t=12976

http://www.reduser.net/forum/showthread.php?t=13159

http://205.234.134.160/forum/showthread.php?t=13168
 
Forgive my ignorance if I'm way off. Would combining the 3 or so photos with different exposures be off if the subject is not static (like different cloud positions and the trees blowing in the wind)?
 
Forgive my ignorance if I'm way off. Would combining the 3 or so photos with different exposures be off if the subject is not static (like different cloud positions and the trees blowing in the wind)?

yes, you're correct, that is the problem with doing the exposures at different times of the day, even if that difference is only 10-15 seconds. For a really good hdr the images should be taken as close to the same time as possible. Usually withing a second or two. That extra time ruins hdr images with a lot of movement.
 
Yes, sure, you'll get strange side effects with moving objects. Photomatix can deal to some extent with the resulting "ghosting", but this will be your biggest problem nevertheless if you want to do HDR in motion. The perfect solution would probably be two or three Reds shooting in pefect sync, aligned in one axis by half-mirrors – a pretty complex setup.

Jeez, we've been doing some tests even with three Canon HV-20s side by side, since they are so small you can get away with it from a certain distance.

BTW, from my years long experience with HDR in still photography and Photomatix you can go more radical with the exposure difference, at least 1.5 stops or even 2.

Next, we could say that HDR is everything where you squeeze a higher latitude from the source into a representation with less latitude, it's not really defined how much more – just like anything with more pixels than SD is HD ;-)

So, considering the pretty good latitude of the Red even in build 15, you might want to try just two shoots about 2 stops apart, could be enough already for landscape and give less of a headache than three. After all, HDR is just about capturing all the tonal values in the scene, check your histograms.

Have fun, I'm looking forward to you results!

Uli
 
Yes, sure, you'll get strange side effects with moving objects. Photomatix can deal to some extent with the resulting "ghosting", but this will be your biggest problem nevertheless if you want to do HDR in motion. The perfect solution would probably be two or three Reds shooting in pefect sync, aligned in one axis by half-mirrors – a pretty complex setup.

It might be possible to partially solve in firmware. If you could shoot at twice the final frame rate, with one "full sector" frame followed by one short exposure frame, it should be possible to get rid of most motion problems in software. This could perhaps be done in some future firmware version.
 
  • Thread starter
  • Banned
  • #9
Thanks all , About Perfect Setup
The perfect solution would probably be two or three Reds shooting in pefect sync, aligned in one axis by half-mirrors – a pretty complex setup.

ok i have 3 Reds but what about aligned in one axis by half mirror :) Did some body nows some ready solutions , 3D Ring from PS Techniknics , can not help me because it will not shoot in 1 axis. Sow if some body can help me with chart of this eqv . Thanks.
Also i will test with only 2 cameras , Sync is not problem problem is to shoot same in 1 axis.
Thanks.
 
I think what biba is refering to is, imagine a rig, similar to stereo rig of pstechnicks, so what that rig does is you have to cameras looking at the same image with centeres of lenses very close to each other for desired stereo effect. is it possible to set similar thing up so that using a mirror both cameras are aligned exactly to each other. this way, if we add nd filter on one of the cameras, we would get HDR. why not try it?

but in the obove example on top, what biba posted is that it was not shot with 2 cameras, or with different exposers. what biba did was took one shot, then in redcine added -1 and +2 on exposure and exported all 3 images, then he combined and here is the result. it is pretty amazing. I think if I knew of any software like photomatrix, but which would work on sequence of images, I would do all my RED footage like that. it would take time but makes sense for me. what do you think?
 
That's exactly what I was thinking, Levan! ^_^

HDR via REDCINE. Wow. I can see this kind of effect applying to any kind of film. The only problem is having to work with each and every frame. Not too bad in practice if you're used to it but definitely a chore nonetheless.

I'm still relatively new to this whole HDR imaging thing but would it be possible to simply do three video files and combine them versus doing three stills and going frame by frame?
 
Using more than one camera..

Using more than one camera..

ok i have 3 Reds but what about aligned in one axis by half mirror :)

For three cameras you would use two Pellicle Beamsplitters and a field lens so that you can put a single wide angle lens in front. A three camera setup would be large and require a crane or dolly to move around.

See the drawings I made for two camera setups, the 3 camera is the same, you just expand the idea and adjust the total path length to be the same for all three cameras.

http://www.reduser.net/forum/showthread.php?t=14479&page=5

Doing HDR from the 12bits in a single exposure of the RED ONE (tm) should not require making three images and fusing them, since you should be able to bend the curves to get the same result, just about. Making three images from a single exposure does give you control of the curves and saturation for three versions of the same frame, that might be hard to do without dividing the image into three images and then back into one image. Does the TIFF hold all the camera 12 bits no mater what EI you use? Is the TIFF linear to the camera's 12 bits, or does the RED software do a curve on the raw sensor data when the EI is changed? At what EI is the TIFF linear to the sensor raw 12 bits?

You should be able to hold most of the highlight and shadow detail in the 12bit camera data by using a S curve to go to 8 ot 10 bits for display.

12bits in the camera is not quite enough to equal pull processed color negative film, if you use two exposures +/-stops and get 16bits in the fusion you have less need to underexpose and use fill lighting.

I was suggesting making static HDR tests with the RED ONE (tm) to see if there are any issues with its sensor and circuit that would make building a two or three camera fixture a wast of money. You can find out if the sensor will flare or there will be overload issues without spending money on something that would not give satisfactory results.

For motion control, the registration issues would be the same as matte work or other multi-exposure.
 
If you want to have these funny colors, often associated with the HDR you can use Photomatix plug-in for PS
http://www.hdrsoft.com/download.html#plugin

REDRAW lends itself very well for the following
http://www.adobe.com/designcenter/photoshop/articles/phscs2at_pixelblend.html
though I was using only AE so far to do it.
Looks like much better way to do it via calls to PS.
I hate the PS scripts though.

With AE and PS script you can process long clips using above mentioned techniques.
It will take some time though to render.

All this will not replace two frames exposed 4 stops apart with double the speed of target material.

Good find Jack!
 
I played around quite a bit with this (exposure method) and if you go 320 you can pull some extra range for sure. As an added thing to do, you can also seperate your blacks on the lower ranges and rather than crushing, isolate them and add gaugasion blur to remove noise. This is limited in dark scenes but should work well when there is less shadow. That way you can focus towards your highlight detail a little (say test at 200 asa) You may get an extra few stops this way but your tonal ranges will really give the perception a run for its money as an added bonus. (It's a lot of work)

If you use two different nd's on a scene it really will only be aceptable if there is a) no camera movement, b) sunny day (no clouds) and c) no wind (or things that can be blown by wind and d) nothing moving in the composition. There's a difference between doing single frame tiffs and moving images.

Using in camera software/firmare where two different exposures are taken sequentialy will have the same problems as above but not as drastic. You still can't have pans or a subject moving through the shot and is therfore very limited as well. Also you are dealing with metadata so really is no different from doing in red alert/ redcine etc.. The only way to really get around this would be to shoot 50 or so frames per second and have a mechanism that is genlocked that either a) changes f stop on lens every second frame or throws a nd filter in front of the lens every second frame. This method is still limited from a proffesional standpoint because slight ghosting artifacts will be introduced to moving subjects or in camera movements (where not all information is being recorded because the moving subject covers it)

Some kind of splitting or mirror setup is where it is at and two or more cameras is the direction to aim for..
 
The only way to really get around this would be to shoot 50 or so frames per second and have a mechanism that is genlocked that either a) changes f stop on lens every second frame or throws a nd filter in front of the lens every second frame. This method is still limited from a proffesional standpoint because slight ghosting artifacts will be introduced to moving subjects or in camera movements (where not all information is being recorded because the moving subject covers it)

Some kind of splitting or mirror setup is where it is at and two or more cameras is the direction to aim for..

No ghosting is noticeable. I played with this already.
You have just blur. You have to help the darker frame a bit though by adding blur to it.
Brighter frame has enough blur so don't touch it, just do Highlight HDR compounding.
Sure, you have to do 48fps for 24 final or 60fps for 30 final.
Thus you will be limited to 2K only.
Slow scenes with people just walking, nothing running you can do at 24fps for 24 final.
 
Levan we have no problems with sequence . Because in Photomatrix you can youse Banch HDR mode , and just select folders with same names of sequence files. You need to render out to 3 dif folders same in - out and it is working. We have some sequence footage from RAW to HDR.
I can not found some ready solution for 2 Cameras , to shoot on same time.
If some body now some solution like stereo ring , please link it .
Thanks.
 
Thanks for posting this Biba.

I had totally forgot about my plans to do exactly the same thing.

http://www.dvinfo.net/conf/showthread.php?t=75342

If you have some footage with sky or some footage ready for hdr. send it to me . I will post some R3D files for all ,
because how to say i am not strong in HDR , sow may be some body we get better effect from same footage.
Today i want to shoot sky , but whether is fine :) Sow i will shoot
sunrise .
 
Also some info about software , and question . I can wotk with 4K Tiff files in Photomatix . But i can work only with 8 Bit , 16 bit not support.
Ok question is :) Did i need to render out 4k , because i can do same with 2K or HD , sow will i get better result with 4K . I thing that it will be same , because all formats will have 8 bit. And resolution dose not have big deal in HDR . May be i am wrong ?
Thanks.
 
Back
Top