Let's Talk

[ByronSanto]

I noticed that switching tabs is recorded into history. Shouldn't only actual edits be recorded into history?

Byron Santo

[HarBal]

Switching tabs is included because edits are contextual in HB3. If for example, you switch to histogram view change the dynamics nodes, switch back into spectrum view change the filter and so on, now press undo a few times, if you are in spectrum view you would not see your histogram view changes undo. You'd press undo and see nothing happen and think it wasn't working. It would be confusing to say the least.

That is why it is necessary to include anything that changes the editing context in the undo stack. HB2.3 could get away with not doing this because it only ever had undo on filter edits in one context so you would always see the undo. Not so for HB3.

cheers,


Paavo.

[ByronSanto]

Switching tabs is included because edits are contextual in HB3. If for example, you switch to histogram view change the dynamics nodes, switch back into spectrum view change the filter and so on, now press undo a few times, if you are in spectrum view you would not see your histogram view changes undo. You'd press undo and see nothing happen and think it wasn't working. It would be confusing to say the least.

That is why it is necessary to include anything that changes the editing context in the undo stack. HB2.3 could get away with not doing this because it only ever had undo on filter edits in one context so you would always see the undo. Not so for HB3.

cheers,


Paavo.

Thanks Paavo!

At least now I know why it's doing.

[ByronSanto]

I understand how to create the track segment, create a filter response and assign it to a specfic track segment. I have even gone as far as creating extreme filter responses for the first segment and used a blank filter response in the second segment. When I play back, I hear the first track segment filter response and see it with the Realtime Power Plot. When playback reaches the second track segment filter response I can see the Realtime Power Plot change but my audio is still using the first track segment filter response. I rechecked the track segments filter response to ensure I did not make a mistake and the outer filter is blank as well and the both EQ buttons are pressed. Am I doing something wrong or have I encountered a bug?????

Also, I totally get the histogram. Great concept! I guess using histograms for my photo and video editing paid off. ha ha .....
Are there any shortcut keys that would restrict the movement of the nodes to horizontal or vertical instead of both.

[HarBal]

The code required to manage multiple filters is rather complex so I wouldn't be surprised if you've found a bug. On the other hand, I'm guessing it will probably be a difficult one to reproduce. Can you easily reproduce it with any track? If so, can you tell me the steps you take.

One thing to try is to select the outer filter and adjust the air. This will force a refresh of the entire realised filter. If you do so do you now hear the right response?

I'm not sure exactly what you're after with the movement restricting shortcuts. Vertical position of a node has no effect on the transfer characteristic. Movement in that direction is allowed simply so you can arrange the nodes in a manner that you find most pleasing to read.

cheers,


Paavo.

[ByronSanto]

I'm not sure exactly what you're after with the movement restricting shortcuts. Vertical position of a node has no effect on the transfer characteristic. Movement in that direction is allowed simply so you can arrange the nodes in a manner that you find most pleasing to read.

cheers,


Paavo.

Hi Paavo

The above question is strictly workflow related. It can be used for ease of editing and to maintain preciseness. Just consider this, I need to move it Verticaly up but I DO NOT want to change any of the node values. Currently that will take some time but if there was a horizontal lock it would simply be a side up with no watsed time readjusting the node values

[HarBal]

Oh yeah. I see what you mean!

I'm blind to the obvious and thanks for the good suggestion. I'll try and think up some key combination to achieve that. Any suggestions on one?

thanks,


Paavo.

[ByronSanto]

Oh yeah. I see what you mean!

I'm blind to the obvious and thanks for the good suggestion. I'll try and think up some key combination to achieve that. Any suggestions on one?

thanks,


Paavo.

Vertically maybe "Shift" + Left Mouse
Horizontally mabye "CTRL" + Left Mouse

I also thought it might be cool to right click on the node to enter a value as well

[ByronSanto]

The code required to manage multiple filters is rather complex so I wouldn't be surprised if you've found a bug. On the other hand, I'm guessing it will probably be a difficult one to reproduce. Can you easily reproduce it with any track? If so, can you tell me the steps you take.

One thing to try is to select the outer filter and adjust the air. This will force a refresh of the entire realised filter. If you do so do you now hear the right response?

cheers,
Paavo.

I can reproduce it with every track and on multiple sessions. I tried the outer filter air adjust and nothing happened. I even pressed the AIR and EQ buttons multiple times. The screen blinked when I did that.

My steps
1) Create splits in track by selecting the location with the cursor, pressing and holding the left mouse button then CTRL+S
2) Select the track segment by left mouse clicking on the segment located at the bottom bar of the time line
3) Time line then auto zooms in to the track segment
4) Create filter responce for selected track segment
5) Press the "double arrow" button to Zoom Out
6) Press the "R" and hold the left mouse button down while I drag the +R to the track segment. This copies the filter responce to the selected track segment
7) Repeat steps 2 - 6 till completed

When I play the track both EQ buttons are pressed I have only had time to test this using the mid filter.

Upon selecting a new track segment I have even tried pressing the "Create New Filter" button and selecing create new filter for selected segment. I have also tried playing the track while in "Track Segment Zoom In" mode

[HarBal]

Hi Byron,

You're absolutely correct. I have appeared to have broken the filter switching somehow in a recent change. Thanks for that! That explains why the demo filters I'd been working on don't sound as good as they used to in a previous revision! I need to improve my internal testing I guess!

Funnily enough, if you cue to the new segment during playback by clicking on the timeline the filter changes but if left to change when going from one segment to the next during playback it doesn't occur.

One point though in you list of instructions on reproducing the issue. It may have been a typo but in step 6 you seem to imply that selecting a segment reference copies the response from one segment to the next. It doesn't but you can copy the response using the new filter dialog assuming you have a segment reference selected.

When I've sorted this bug out I'll send out a revision to all beta testers as it's a pretty serious issue. It may take a bit of time though as I'm caught in the middle of another more complex issue to fix. I'll try and get it out in the next few days.

cheers,


Paavo.

[ByronSanto]

The average power plot is louder then the peak power plot. This is the first time I have seen this happen. Spectrums are correct

[HarBal]

A higher peak value in the average power histogram in than in the peak power histogram does not mean the average power is greater than the peak power and it can happen. The y-values in the histogram correspond to probability not power.

Having a high peak probability just means that the average power envelope hovers around the same level for much of the time.

A typical case of where the peak values may have more variance than average would be for a track where the performers play with constant intensity (say forte). In such case the average power will be close to constant but the transient material isn't so you'll end up with more variance (greater spread) in peak levels than in average.

cheers,


Paavo.

[ByronSanto]

I was doing an experment on a really bad............. mixdown. I was taking a 100% visual approach. Speakers turned off. I corrected the extremely bad dips in the MID. Switched to the SIDE and corrected a few more problems. Gave a listen and it was better then the original but not great considering that I was 100% visual. Turned the speakers off and decided to check LEFT/RIGHT. The bad frequency dip that was present in M/S was present in L/R therefore since I was 100% visual I decided to correct it. Bad mistake. Now I was boosting the frequencies. Then it hit me, maybe there should be a reanalysis option for either the L/R or M/S once the user has created a filter(s). The frequency reanalysis would facture in the existing filters, be selected by the user and would only be limited to L/R or M/S.

If it would not have been for the bad mixdown I was working on I would have never noticed this.

[HarBal]

Hello Byron,

You've lost me a bit there.

HarBal 3 accounts for as much as it can in the predicted spectrum post filtering. The bits it can't show you is the effect of mid and side filters on left and right or indeed, the effect of left and right filters on mid side.

The reason is why that is the case is simple. The analysis file contains reduced analysis information in time (basically 1/12th octave binned spectrum estimates) without any phase information so it is not possible to determine the mid and side spectrum's from the left and right spectrum's. To be able to do that requires storing a lot more information than is practical to do.

I suspect what you may have done is seen a large dip in the spectrum in left and right and the same in mid and created corrective filters for both. Don't! The reason being is that you will have corrected for the problem twice and hence undone your correction. You should only work in one or the other to correct spectrum imbalance but not both, simply because you won't know the effect of the change in spectrum of one domain in the other and visa versa. Personally, the left / right filters are something you typically should very rarely use anyway. They are there for exceptional circumstances rather than the general case, and because we can easily implement it so why not.

cheers,


Paavo.

[ByronSanto]

Hello Byron,

You've lost me a bit there.

HarBal 3 accounts for as much as it can in the predicted spectrum post filtering. The bits it can't show you is the effect of mid and side filters on left and right or indeed, the effect of left and right filters on mid side.

The reason is why that is the case is simple. The analysis file contains reduced analysis information in time (basically 1/12th octave binned spectrum estimates) without any phase information so it is not possible to determine the mid and side spectrum's from the left and right spectrum's. To be able to do that requires storing a lot more information than is practical to do.

I suspect what you may have done is seen a large dip in the spectrum in left and right and the same in mid and created corrective filters for both. Don't! The reason being is that you will have corrected for the problem twice and hence undone your correction. You should only work in one or the other to correct spectrum imbalance but not both, simply because you won't know the effect of the change in spectrum of one domain in the other and visa versa. Personally, the left / right filters are something you typically should very rarely use anyway. They are there for exceptional circumstances rather than the general case, and because we can easily implement it so why not.

cheers,


Paavo.

Paavo

I wondering if you should have a option to select either L/R or M/S mode then. Then the user would be forced to select only two filters and not make the type of mistake I did