Wow, I just found my longer reply! I had the wrong window open. Doh!
Quote:
In Standard Gamma the 'curve' is linear, ie straight (if we assume there is no knee). 100% input equals 100% output. The thing is though that the CCD is capable of capturing, in the 1/2" cameras, a 400% more signal over and above the maximum recordable output highlight level. So, you have to compress the highlights in order to cram that extra signal into the recordable picture. The result of that is that you are recording a higher range of shades within the same recordable signal area, and so graduations might be coarser, even though you are seeing more actual highlight range. Hence limited tonal range. If you push things too far in post to try and put more punch into the picture, you may end up seeing the limitations. A 10-bit recording format would be much better at handing such curves. |
I'd add to Simon's definition of contrast, that what most people consider "contrasty" is greater range between blacks and highs, while Simon's high contrast is about the resolution of midtones in between. Hence: cine gammas are more contrasty, and standard gammas - more punchy.
This is consistent with how the various gammas are (briefly) described in the EX1 manual; counter-intuitive at the first sight. |
I'll also add that what i know about the technicalities of all of this is what I have gained by picking engineers minds. If they were members of these forums they'd be able to describe what I am trying to put into words with far more accuracy. But the caveat would be that we'd all be suffering from a far worse headache! ;-)
Alister, feel free to chime in here! |
Quote:
Conversely, if one of the Cines is used and it fills the histogram from right to left, the recordable contrast ratio is already maxed out, and switching to STD can not increase it (but will blow the ends off, or at least one, depending on settings). Side note: I am finding the EX-1 files to be nicely "curve-able" without noticable tonal degradation (so long as a reasonably bright exposure was achieved initally). |
Quote:
You have an empty glass jar of 100mm height, and you have to fit a 140mm high piece of flexible wire into it. The top of the wire represents the highest white possible that the CCD's can capture, while the low point of the wire represents the blackest black. The top of the jar is the maximum recordable white level. How do you fit the wire into the jar? The Cinegammas, particularly 1 and 2 will output a maximum ire below 100. But this is not the same as maximizing the contrast level on a TV set, or in post using the Levels adjustment. Quote:
|
You can see this std vs cine demonstrated by setting up a shot with std1 and then without touching anything switch to Cine1 and you will see just how much you can crank exposure to get a lot more at the top.
|
[QUOTE=Simon Wyndham;836308]Hence we are talking about two separate things. Ignore your definition of contrast. The contrast I am talking about is the contrast range or luminance range that is *captured* from the CCD's and then compressed into the final signal and *not* the final effect of a histogram or 'look' of the picture in terms of punchiness.QUOTE]
Hi Simon, As I understand it, the "pre-compression" contrast range that is read by the CCD's would be the same in both instances and as you said, would be based on the technical capabilities of the system. The curve (or lack thereof) that is applied forms the captured image. In terms of a user choosing a Gamma setting based on contrast alone, the pre-compression contrast is of little use, as there is no going back. The wire analogy defines value remapping and also can not be undone. If you meant to say that the Cines take the existing contrast level at the sensor and compress (or remap) them into a lower contrast range then we are saying the same thing. |
Quote:
I will try to add more grabs this week, changing gamma, knee (for std gamma curves), etc. Check this here : http://www.lecentre.net/blog/pmw-ex1/gamma |
Good one Sebastien. You saved me the trouble as I was going to write a similar article.
With the adjustments available on the EX1 and the F300 series, I am surprised that a SAW signal isn't in the standard menus as it is essential for setting up your own gamma curve settings. |
Very interesting guys, and thanks for the post.
I have to play village idiot again and admit to not understanding the fundamentals here. If I have it correctly, these are representations of what is happening to a linear 0-255 gradient when each of the gamma setings is applied. Is that correct? If so, in (for example Cine 1) why is every value from 207-255 being mapped to a flat 207? I thought I understood your wire analogy, but did not imagine the wire being bent flat at both ends. I thought we were talking about a smooth rolloff as is depicted by the curves in the user manual, but certainly not a long flat spot from 207 on. Well...I can not understand what value there would be to that, but then again, I have always found pre-processing curves to be a mental puzzle. As you can see, I need to be set straight. HHHEEELLLPPPP! |
Quote:
I will update the article with 0 gain pictures. Except that, the theory is true. |
This all raises another question which has been nagging at me. It seems like there would be a lot of value to some curves with the overall range reduced similarly to the Cine curves, but much more linear in nature (especially less compression of the whites). In a lot of cases (see the exmaple titled watermen I posted here http://dvinfo.net/conf/showthread.ph...498#post843498) I am seeing situations where I need to protect my whites, and thus need dynamic limitation (for lack of a better term), and have been careful not to overexpose any channel over 230 for safety (and potential user error), but the whites still appear blocked up as in overexposure, due to such a broad range of whites being compressed to such a narrow range. Also, we are pushing the blacks more than might be needed (I am typically backing of black gamma in post). Has anyone taken a crack at a PP (or group of PP's) like this?
|
Quote:
|
Quote:
I think you will find most are backing of blacks in post or in the camera with negative settings for master black and black gamma. To get a good spread of whites use the histogram and push exposure a bit. I started off preaching the wisdom of under exposing and Bill almost sh*t down my neck but with experimentation I think he is half right. I say only half because from my experiments when the 100 zebra shows up (Bill's point) you are still a stop or two underexposed. Try going further for which you will need watch the histogram so as to stop pushing it short of blowing. |
Quote:
Thanks Michael, So many confusing terms. Negative settings for Master black sets the black point below zero. I think that may have been the wrong way to start for my type of shooting. It did deepen the lower quartile darks that I was concerned with, but with the additional consequence of driving all near blacks through the floor. We can only "Gray those back up" in post, but with noise of course. I am envisioning a curve with black level at Zero and black gamma reduced. Whites starting at ranges from 100 to 80 with a lower and less drastic knee on each. Your last point is the tricky one, because we are all right, kind of. If the curve was more linear, pushing to the right would act more like RAW. The test above (although accidentally) shows the current problem with that with our current curves. One stop over, and due to the nature of the curve, everything from 255 to 207 gets mapped to a single value. Even in a case of proper exposure they are getting mapped to such a narrow and similar range, that in many cases they will "read" as blocked up or blown. |
I did a bit of playing around with SAW in an attempt to better familiarize myself with what our camera is doing. I should note, that this is not any kind of negative critque of the camera, just one user trying to figure out some best practices.
Regardless of the camera's gain setting, a flat spot exists in each of the available gamma settings, indicating to me that all values above a specified point are being mapped to a single value. Knee (on or off) changes the curves transition into this flat point, but does not alter it. Why would it be valuable to map (for instance) any incoming value of over 200 to an output value of 200? Apologies, I feel like I am missing something obvious, but what is eluding me. |
Ok I'm out of this thread. I think I've got all I need now and I know where I was going wrong.
You can not set a few PP and use them only - you must adjust some paint settings like master black on every shot. I can eliminate any need to manipulate the blacks in post. I think the CINE gammas despite the criticism of the white range handle the white roll offs fine as long as you use CINE1 for very bright scenes and keep Cine4 for the gentler light. My remaining decision not yet made is HiSat or Cinema. HiSat (compared to Cinema) has a yellow ting to me but I have not yet compared with Bill's color shifts |
Quote:
Whatever, you will ever have differences between what you see (and what you may get in 10 bits SDI) and what you record, 8bits + codec. |
Quote:
The "flat spot" is present in all PP's and is very relevant to me, because it explains a lot about how whites are being handled. I owe you guys a big thanks, because I had no idea that the SAW function was present and you are absolutely correct that it is imperative for adjusting these PP's with any accuracy. |
True Color with Letus EX1
I have followed this thread for a while. I have used and loved the presets developed over time. Thanks for sharing all the work.
I am curious if anyone has formulated a true color preset with a Letus and EX1. I realize this is somewhat imprecise as it depends on the lens used, but it would still be helpful. I know Phillip is now shooting with both a Letus EX1 and a "naked" EX1. I am going to be doing the same. I have 2 EX1's, one with a Letus one without. I use primarily my Zeiss 85mm 1.4 ZF lens. I would love to find a preset that matches the base EX1. Phillip, do you have any recommendations? I will have a break in productions over the next few weeks, so I will try to do some work on my own, but would appreciate any advice. Thanks, Adam Simpson |
Quick question regarding crushing or stretching.
If my black is set at 0 and i go down to minus whatever,say -10, is that crushing or stretching ? Thanks. Paul. |
Quote:
|
Thanks Piotr.
That's what i thought. |
FWIW...
Studying the SAW display in HDRack's waveform monitor, it is apparent to me that a gain setting of -3dB is NOT equivalent to the "real" 0. A GAIN setting of -3dB limits the superwhite IRE to a value of ~90 IRE. Only the 0dB GAIN with 3200K for a white balance yields a neutral, full scale curve. Along with this, I will also note that my previous BLACK settings seem to be causing noise in the shadows, indicating the shadows are being crushed. With data from the SAW that Randy provided, the min setting for BLACK I now recommend is -3 or -4. |
Quote:
|
Thanks Bill nice to know that my eye was happier with -2 now I know why.
|
Quote:
|
Quote:
|
Quote:
|
log data
ive been thinking about this, i dont think it has been posted about, but given the picture profile controls available in the EX1, i wonder if it would be possible to create a log-like picture profile so that the information recorded by the camera does a better job of retaining the 14(?)bits initially captured from the CMOS. 8bit log data could be extrapolated much more accurately to 10bit linear data. of course this would require a specific workflow to deal with the LUT required to see the picture correctly, but it could help a lot in maintaining latitude and detail. might take some trial and error to find exactly the right settings, but the camera is so configurable that it certainly seems possible.
|
Quote:
I should have read who you had quoted more carefully. I only scanned the quote and thought you were commenting on Bill's suggestion of Black at -3 or -4. In the quote you actually posted Micheal Stevens was making a suggestion as to how the SAW Test may have better been conducted. I do not think he was making any general usage suggestion on gain there. Back up a few posts to read the responses to that suggestion. Best, Randy |
This is all very interesting but what does this mean in practicality? Does it mean that the default gain should be zero and that even thought he menu for crushing black master goes to -99 we should limit it at -6? In bright light with CINE1 and gain at -3 I have needed take it down to -12 on may shots to maximise the bottom of the waveform as first recommended by Randy.
|
Quote:
|
I would like to modify something I had said earlier about the Gamma setting in PP's.
I had stated that this setting would be like adjusting a 50 percent middle point on a curve or in levels. I had tried to put this in terms of adjusting gamma in post, or in an image editing program such as photoshop. In those invironments, Gamma pulls at the mid tones, while leaving the black and white points static. Not so in PP's, adjusting Gamma leaves the black point static, but adjusts the white point. It actually appears to me that "White" would be a more consistant name than Gamma. This may be quite common in video PP settings, I do not know, but I wanted to clarify so that my earlier comment would not confuse anyone. |
Quote:
|
Quote:
I am looking at the resultant curve live on the waveform on my monitor as I make the change, but please confirm this for yourself by recording the saw pattern to SXS. I had originally looked at the saw pattern for Cine 1 at a color temp of 3200 and gain at 0 with the only other adjustment being Black pulled to the floor (-4). I then adjusted gamma up and down by about 50 points in either direction. No change to black, only to white. Then, as black was on the floor, I wanted to give it some room to see if it would adjust down, so I raised black off of the floor to 0 (which sets it at 4 IRE). I then repaeted the gammma up gamma down about 50 points in either direction. Same results. Black stays static. Everything stretches from white. After I saw your post, I rechecked by pushing the gamma all the way to plus 99 and negative 99, and at negative 99 with black at 0 (4 IRE) it will lower by 1 IRE (to 3 IRE). At the same time white is moving by 10+ points. I then checked STD1 and the results are very similar with only the smallest motion to black but with largest sweeps to white. If there is interest, I will be happy to snap some shots later. |
How do I record the saw pattern?
|
Quote:
First, get into the hidden menu options by pressing (at the same time) cancel, select (the rotating button) and menu. You will reach the usual menu, plus 3 items. Select saw pattern and turn it on. Then get out of the menu option and into the pp menu. You can record live to SXS while you make the changes. |
Quote:
Also I am videoing a wedding on Monday. The church is quite dimly lit, but the aisle shot is likely to be backlit from the main windows at the front of church. I would like to have more detail in the blacks so can brighten them if necessary in post. I will probably need to use several picture profiles in the church. 1) Aisle bright backlit shot in dim church 2) Stage shot, under a mixture of white stage spots (unfortunately not evenly dispersed lighting) 3) Signing of the register - dimly lit. Is cine4 better in bringing up more contrasts in the blacks or would std1 with a modified PP be better. I would like some practical answers please! Thanks |
Ex1 vs. XL H1
I just went on a shoot for promotional video. The director is a 35mm film purist who was doing the shoot for charity and had to use low end HD (my EX1 and another XL H1). When he asked which should be used for A camera and which should be used for B I said use the XL H1 since it had a wide angle lens...but by the end of the night I feel like my camera had turned into the A camera along the way...thanks to the film curve pp...the closeups on the EX1 were just more impressive...yayy film curve pp...thanks for posting this stuff dudes and dudettes...
:D |
All times are GMT -6. The time now is 10:20 AM. |
DV Info Net -- Real Names, Real People, Real Info!
1998-2024 The Digital Video Information Network