[Michael Mccune]

Marc; Good news regarding this mystery. And like many good mysteries, the resolution was entirely accidental.

While cleaning up stuff I found that the cache was–based on my limited experience–huge, about 1 TB. True, I had experimented with various cache modes. And without fully researching the possible consequences, I just deleted the entire contents of the cache folder.

No only did Resolve seem to be noticeably quicker but the mask issue you helped with now seems to work as we expect!

And further, it seems that the video “defect” in the original 4K ProRes 4:2:2 recording also now seems normal on playback. This “defect” now seems gone. This was about 1 1/2 seconds where a portion of the frame area dropped in luma value, with unusual waveforms seeming to show digital comb defects dropping these pixel-wide slices of the image to zero.

Ever heard of such a thing?? That is, short defects in video, perhaps introduced during editing or color work???

And, as far as you know, Marc, does the cache normally affect the display, such that a defect could be introduced there and continue to affect subsequent playback and output until cleared???

Thanks again for all you help, Marc.
Mike

[Michael Mccune]

Thanks, Marc; Very encouraging to hear; surely there has to be be a way.

But I confused the issue by using the work “track” with regard to matching color changes at each keyframe.

Yes, I am using manual node keyframes, not using the tracking tool to mask a moving visual element. That is, I manually create each keyframe, move the mask as needed and apply some color (or other level or hsl) change.

But the color changes then apply to all the manually created keyframes. So, while the mask itself does animate its position with corresponding keyframes, the color value does not change at each keyframe.

A simple example using just two keyframes: with a simple static visual background (and no tracking) I create a simple circle on the left side of the frame and manually create a keyframe at that position. I set some color value, say gain at 110%. As expected, the center of the circle is slightly brighter.

I move the current time indicator along the timeline, say, one minute and create a second keyframe and move the circle mask to the center of the screen. The centered mask now shows its slightly brighter gain of 110%, inherited from the first keyframed mask color setting. I raise the gain to 130% to distinguish this keyframed center mask value from the first keyframed mask color value. This second mask at the center of the frame then shows, as expected, gain and brightness at 130%.

But here is the problem: go back to the first keyframed mask circle on the left side of the frame. What do we see? Is the left side circle brightness showing at its initial value of 110% or at the 130% of the second keyframe? Answer: The first mask brightness is showing at 130% just as it is for the second mask.

In other words, any color adjustment applied to one keyframed mask is also applied to all the other keyframed masks.

Nuts. Must be missing something very simple.

Clue? I have used tracking a lot in other nodes in this clip. But I created a separate node on this same clip node tree for these keyframed tests.

Thanks,
Mike

[Michael Mccune]

Thanks for the insights which are well taken, David. (Also, interesting article in Tom’s Hardware about the trajectory of SSDs now with 2 and 3 layers for increasing size and the corresponding plateau of hard drive density. They say another two or three years and we may see rough cost parity along with SSDs having more than 8-10 TB capacity–just add more chips– which now is about the limit of hard drives.)

The 4K source I have is Sony recorded onto Shogun ProRes 4:2:2. You would know better, but ProRes seems to be well-liked and very workable for editors.

I believe your suggestion regarding the Titan X is also on target.

I appreciate your time, David.
Mike

[Michael Mccune]

At 4K??? H.264 for HD, yes, but for 4K??? And this is on a PC, not a MAC, which can export to ProRes .mov, I understand.

Please detail your PC Delivery options to allow 4K output in a format recognized by the new 4K monitor/TVs, such a mov.

Thanks,
Mike

[Michael Mccune]

Right. Thanks.

I was not clear. This is not an editing viewing monitor question, rather a home theater viewing question.

The idea is 4K playback into a conventional large screen TV (or monitor or projector) from the finished file output from Resolve.

As such, it is played back from a disk, large USB 3 drive or home theater micro pc (without a dedicated BMD card).

We know that H.264 with an .mp4 or .mov file wrapper works for standard HD into a monitor/TV/projector.

But does Resolve have a corresponding output format for 4K???

Gotta be a way; all the new camera stuff is touting flavors of 4K.

Thanks,
Mike

[Michael Mccune]

Right. These are changes I have been trying. That is, smaller clip lengths and an HD timeline, not 4k. Also, it seems that the latest 12.01 version improves memory management. Memory seems to climb–viewed using Windows Process Explorer–until the entire 32GB is used. So I save before that point and restart Resolve. Interesting in that this still takes a minute or two after Resolve quits for the “committed” ram to fall back to a normal level of about 6-8 GB.

Overall seems to be working better.

Question: for this simple case of tracking a single face in an interview format–for the simple purpose of lifting the face light levels slightly–would you use, as I have been using, a simple elliptical shape with feathering or would you use a more precise but certainly more demanding bezier-based shape??? Do you find that one tends to track better for this simple purpose??? The face of the subject does move from full left profile to full right profile, which is when the tracking needs a little help.

There seem to be few tutorials on Resolve tracking. Even Lynda.com. A couple of editing training sites may include this subject in their videos but that is not clear from their descriptions.

Thanks again,
Mike

[Michael Mccune]

Very helpful and, as is typical of technical mysteries, the answer is that an *implicit* keyframe exists, the first frame itself, a static keyframe, which holds until the next keyframe.

Probably the same at the end frame, an implicit static ending keyframe.

Clever.

Thanks
Mike

[Michael Mccune]

That is a help.

I predict that the “Deliver” page will get a lot of attention in the non-beta version update.

This last page perhaps received the least attention as crunch time approached.

I believe NAB rules say something about new product ship dates–within four months???

Mike

[Michael Mccune]

Thanks. Is there a detailed “reference” resource for keyframes???

Please clarify usage of different keyframe types with respect to interpolation. Dynamic types will–I believe–interpolate in both directions, as indicated by their left and right facing arrows.

So a sequence could be made exclusively of many of them– all dynamics and no statics– right?

Or, should the head frame and the tail frame be capped with a static type???

Will a dynamic keyframe that is, say, one minute from the start interpolate *back* to the first frame, which would not have a keyframe of any kind???.

Or, will that first dynamic assert its settings back to the start at 100% without interpolation, perhaps because there is no other frame with which to interpolate???

Thanks,
Mike

[Michael Mccune]

I see the next and previous keyframe shortcuts given in the menu, which are [ and ].

How about a tear-off window for keyframes, similar to the waveform window that allows users to dock it on a second screen???

The obvious advantage is re-sizing for easier visibility and closer inspection, which the tiny keyframe window does not allow.

True, one can expand the keyframe window but at the expense of immediate access to the color controls that are likely needed.

Or, perhaps someone has a simple workaround, or perhaps I am missing something.

Any future in this for upcoming releases???

Thanks,
Mike