Enhancement Request for Next Release

[whoareyou]

Hi Bernt,
For next release @juicehifi , will you please consider saving changes made to the graph's UI settings; Background colors, series colors, etc. ? Better yet, maybe a subset of these colors can be user configurable via settings?

I literally cannot see some of the lines on graphs and always need to change them while working on graphs. Unfortunately, these changes are not sticky, so I'm constantly resetting these so I can see some of the graph's info.
Thanks

 

[Omid]

Here are a couple of other enhancements that would be awesome (pie in the sky, probably not a priority for Bernt):

The ability to run a correction repeatedly while varying a variable. For example once you run a correction filter, you would instruct Audiolense to run the calculations again while varying the crossover frequency from 40 Hz to 80Hz in increments of 2 Hz for example (Saving 20 different versions of the filter, So you can choose the best frequency &step ). Or vary the window length from 4 to 12 cycles. Basically to automate the trial and error process. There are so many variables, that I probably get tired before landing on the optimal choice. I wouldn’t mind letting the computer work for two days, and spit out a large number of responses that I can rapidly scroll through at the end to choose the best.

another useful idea would be for audiolense to apply a numeric score to the response obtained, based on the flatness of the frequency response, the degree of deviation from the target step response, the degree of coherence between right and left channel. I realize such a score would be a bit arbitrary, but it would make it quicker to sort between a variety of responses.

 

[juicehifi]

Hi Bernt,
For next release @juicehifi , will you please consider saving changes made to the graph's UI settings; Background colors, series colors, etc. ? Better yet, maybe a subset of these colors can be user configurable via settings?

I literally cannot see some of the lines on graphs and always need to change them while working on graphs. Unfortunately, these changes are not sticky, so I'm constantly resetting these so I can see some of the graph's info.
Thanks

Changing the series colors are lot harder than it may seem. The automatic color generation includes a lot of unreadable colors, so I've made my own library with colors that have decent visibility. Furthermore, there are som many different scenarios for what is shown. In short; It's a rabbit hole and I do not plan to go there.

I may be able to do something about choosing back ground color, though.

 

[juicehifi]

Here are a couple of other enhancements that would be awesome (pie in the sky, probably not a priority for Bernt):

The ability to run a correction repeatedly while varying a variable. For example once you run a correction filter, you would instruct Audiolense to run the calculations again while varying the crossover frequency from 40 Hz to 80Hz in increments of 2 Hz for example (Saving 20 different versions of the filter, So you can choose the best frequency &step ). Or vary the window length from 4 to 12 cycles. Basically to automate the trial and error process. There are so many variables, that I probably get tired before landing on the optimal choice. I wouldn’t mind letting the computer work for two days, and spit out a large number of responses that I can rapidly scroll through at the end to choose the best.

another useful idea would be for audiolense to apply a numeric score to the response obtained, based on the flatness of the frequency response, the degree of deviation from the target step response, the degree of coherence between right and left channel. I realize such a score would be a bit arbitrary, but it would make it quicker to sort between a variety of responses.

IMO, you do not need to try everything to get the best possible result. Here are a few guidelines:

If you see problems with the frequency response, increase the measurement & correction window, the max dip boost or both.
For the TTD correction you may need to try a few different options to get the best possible result. On the one hand, you want a very clean entrance to the step. On the other hand, you want a decay pattern that has a similar shape as the target. Sometimes there can be a tradeoff between the two, and perhaps you need to listen to decide.

These windows sizes often have less influence than you may think. I only experiment with them when I see problems in the time domain (the step response).

Crossovers are a different topic. Knowledge about the involved drivers are important: Especially their good frequency range. In addition, pay attention at low frequencies, wheere room interaction can lead to poor driver integration. Sometimes a different crossover frequency will work better.

 

[whoareyou]

I may be able to do something about choosing back ground color, though.

Ability to select background color (and persisting the selection somewhere in settings) would be a big improvement, and probably good enough.

Changing the series colors are lot harder than it may seem.

Ok.

The automatic color generation includes a lot of unreadable colors, so I've made my own library with colors that have decent visibility.

Sorry, I'm not following as to where this library exists. Do you mean that every time you work with a graph you manually assign one of your custom colors from some external source?

Thanks

 

[Omid]

IMO, you do not need to try everything to get the best possible result. Here are a few guidelines:

If you see problems with the frequency response, increase the measurement & correction window, the max dip boost or both.
For the TTD correction you may need to try a few different options to get the best possible result. On the one hand, you want a very clean entrance to the step. On the other hand, you want a decay pattern that has a similar shape as the target. Sometimes there can be a tradeoff between the two, and perhaps you need to listen to decide.

These windows sizes often have less influence than you may think. I only experiment with them when I see problems in the time domain (the step response).

Crossovers are a different topic. Knowledge about the involved drivers are important: Especially their good frequency range. In addition, pay attention at low frequencies, wheere room interaction can lead to poor driver integration. Sometimes a different crossover frequency will work better.

Thank you that makes a lot of sense.

 

[juicehifi]

Ability to select background color (and persisting the selection somewhere in settings) would be a big improvement, and probably good enough.


Ok.


Sorry, I'm not following as to where this library exists. Do you mean that every time you work with a graph you manually assign one of your custom colors from some external source?

Thanks

The program assigns a color to each graph.

 

[whoareyou]

The program assigns a color to each graph.

Thanks

 

[moedra]

I may be misunderstanding the issue, and I know you've said you don't want to mess with colors, but it seems to me like all that is needed is a flag in the settings that keeps the current color scheme whenever you change one in a window. It doesn't need to change any colors, or do anything related to generating colors. It just needs to not reset the theme every time something is refreshed, or a parameter resets. The color schemes of the main windows, like frequency, group delay, and impulse actually do stick as long as i don't close Audiolense, so why can't the other color schemes stick? The most frustrating one for me is the measurement noise cleanup window, which resets every time I click the mouse to adjust the marker. I can't see what I'm doing in there and it's very cumbersome to work with because I have to change the theme back to something visible every time I click somewhere just so I can see the change I made. All we need is for the themes to stick where we set them and not reset.

 

[juicehifi]

The color scheme is very difficult to perfect.

There is an automatic solution that generates colors on the fly. But on systems with many speakers and drivers you get color combinations that are virtually impossible to set spart. So I have made a scheme with colors that imo works OK. They may not be up to your standards, but it will be extremely complicated to change this for the better. There are so many scenarios….

And a change is not guaranteed to be an improvement either.

So sorry to say, but this is a low priority at the moment.