Update: BeatLog-II + Beat's Flat

Started 2 months ago | Discussions thread
Joacim Schwartz
Joacim Schwartz New Member • Posts: 6
Re: Update: BeatLog-II + Beat's Flat

beatboxa wrote:

Joacim Schwartz wrote:

beatboxa wrote:

Joacim Schwartz wrote:

beatboxa wrote:

Varuas wrote:

worldaccordingtojim wrote:

Can you share the picture profile?

Beatboxa is one of the best posters here and we are lucky to have him use Z and share his expertise. That being said, it would be great if he uploads these profiles (name them beat-log-II, beat-flat, etc) to google drive or somewhere we can download. I tried replicating his profiles but not sure if I did things right. It would be download his profiles and compare.

Please do consider it, beatboxa!

Appreciate the kind words, and will do.

Give me a bit. I'm still checking some things out, and I'm debating whether the flat (which I've since tuned a bit more) or the log will actually end up being better.

It's unfortunate I won't have the camera for a few days to test with.

I can only agree with Varuas comment, so much great work from Beatboxa! It's a joy every time your name pops up in a thread.

I have been trying to replicate the curve you described in Picture Control Utility but for some reason, my results always have more noise in them compared to something like eoshd L-log. Is this by design or am I doing something wrong? Any update on sharing your profile? I can give you a dropbox link where you could dump it into.

Thanks for the kind words. Believe it or not, "more noise" is actually to be expected!

Remember that the purpose of this profile is not to come close to your final output; but rather to store as much usable data as you can. If you don't plan on grading the footage, don't use this profile because there are much better profiles out there.

Part of this "max data storage" involves boosting shadows quite a bit, with the expectation that you will probably darken some of the shadows later when you grade. And shadows is where the noise lies. So it's not so much that this gives more noise, as that it makes the noise already there more visible, allowing you to see the exposure better. It does this because this is a pretty much unavoidable area--I can always darken this more, but that messes up all of the other shadows while freeing up maybe 5 tones or so.

If you look at the original histograms, you'll see an area on the left where it suddenly goes from black to grey. This point is the noise floor, and it's where the camera cannot discern between its own electronic noise and any signal. But the point right after this is where the useful information begins. You can safely darken this area, while other profiles already start dark, unfortunately taking some shadow tonality along with them.

To use this properly, always make sure you expose as much as you can--expose until you see the zebra highlights. If you are and this is still noisy, the problem is your lighting, not the profile rendering. You can also tryraising your ISO to at least 800 to lower the electronic noise, as long as you don't clip highlights in doing so. I know this may sound counterintuitive, but it's correct.

Post a sample to confirm this is what you're seeing.

I have now tried compared three different profiles; EOSHD Z-Log, my attempt at BeatLogII and BeatLogII with a small boost with the middle slider to 1,15 within PCU2.

Shot with 25p 50/s, ISO800, f4.

I did not frame exactly the same but close enough. What I do like about BeatLogII, compared to Z-log, is the highlights are more represented as it is viewed on the screen. My Zebras/Highlight display is set to 235.

I know that screenshots are not the same as seeing the video playing, but this kind of shows what I see.


Even when correctly exposed, a larger amount of the highlights are not recoverable with Z-LOG.


BeatLOGII + 0,15

When pixel peeping at the shadow/dark areas I find that Z-LOG has more noise at the darkest parts (under the desk) but have a nicer pattern at slightly lighter parts (backside of the chair)



BeatLOG has nicer noise at the darkest parts, but I noticed that the back of the chair looks more blotchy, compared to the other profile.

BeatLOGII + 0,15

I know I'm nitpicking and pixel peeping, but comparing it looks I have done a terrible job re-creating the profile. I get that BeatLOG is not supposed to be as compressed or that getting more DR I need N-LOG or PRRAW.

What are your thoughts on this?

Their Z-Log bears a resemblance to one of my earlier (purely calculated) ones here:

This type of profile is useful in extreme situations where a majority of your scene is underexposed--ie a majority of interesting subjects are in the shadows (such as the scene in that link); but you also have some highlights you don't want to clip--in this case, the sun is rising, but the subjects are in the dark, facing away from the rising sun. But normally, I would avoid this type of profile because most scenes don't fall under this category. And the more room you give shadows, the more you take away from midtones & highlights.

And as I mentioned, that was was purely calculated, which has a flaw: we don't actually start with raw sensor readout. In other words, the processing chain goes like:

  • Raw readout > Nikon Flat > Log picture curve

And the "Nikon Flat" is the curveball. Ideally, we would be able to skip this middle step. A different approach I took in this thread was to actually measure the input vs output (through the use of the gradient) and use that to create a curve. The result is what I think is a generally more useful and cleaner output in a majority of cases.

I would say the real question is how things look in practice after grading rather than before grading. For example, if you look at the Z-Log, there is no black, because even the noise is boosted quite a bit to grey. So what's happening is not that the noise has disappeared, but that it has been boosted and muddled in with the shadows. When you grade, you will inevitably darken these, and the noise may become more apparent again. (Same thing that happens when you shoot raw stills--there is much more tonality than appears on the surface).

So, for example, if you align the darkest shadows back down to black with no other changes, here is the result of ZLog:

And here is BeatLog-II:

Because they gave--or I would even say wasted--a significant amount of tonality for noise. As far as I'm concerned, that's noise anyway, I don't care about it, and it should be black. I think I mentioned this in my original post, but I could have squeezed an extra 10 or so tones out of this by darkening this area, but it was just too difficult in Picture Control 2 (to do properly, with a 2nd pass) for very little gain. If you really want to do this the quick way, slide the leftmost point in Picture Control 2 right a few notches--maybe to about 8--and you'll gain roughly 3% tonality. And 3% tonality is nothing.

Maybe I'll get around to doing this one day in BeatLog-III or something.

Great information Beatboxa! I think I'm just expecting too much out of a Picture Profile, I have been watching too much Sony S-Log footage online and got jealous of the supposed DR, but I will try your example about lifting the left point!

Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow