DSP Robotics Support

billv wrote:What do you think about the "creating a ppq fix in red" for the midi to poly?

Honestly I don't really know anything about that. I'm still green to Flowstone.

The waveforms looking the same may be some kind of coincidence. On my test I capped images of two different notes at the beginning and end and they just happen to look identical, but otherwise my notes are all starting at different positions. Maybe you somehow managed to line up the osc frequency with the tempo perfectly? Or maybe you did something to cause the osc to reset at every note somehow?

I made a post to tech support on the IL forums about why the view in the playlist appears to drift while the view in Edison doesn't, so hopefully we can get an official answer to clear that up. My suspicion is that it's a visual inaccuracy and nothing more, but what do I know?

Perfect Human Interface wrote:Maybe you somehow managed to line up the osc frequency with the tempo perfectly? Or maybe you did something to cause the osc to reset at every note somehow?

Don't know mate, like i said:

billv wrote:Yes. export it. enter notes in fl seq. record it.

That's it. Don't see how where getting differring reults, so really curious as to what your result looked like
Mine was 3 min test.
Taking first screenshot at the first available black host marker.
Another screenshot at the last available black host marker.
Can you re-test and up-load some shots?
This would help me a lot to bridging the gap of understanding and put us both on same page.
PS:
if you get annoyed doing printscreens, I really recommend a small FREE program called
"Gadwin PrintScreen"
http://www.gadwin.com/printscreen/
it takes all the pain away from taking quick shots

Perfect Human Interface wrote:but what do I know

Your help is still very, very welcome. I'm still making this up as I go along anyway...what do i know
Anyway, I'm going to try some simple ruby/ppq midi fix stuff on the Stock VSTi..........

Wow guys, I'm really on a "roll" at the moment.
Attempted fix to Stock VSTi was successfull
Polarity tests for X11 2.01 timer may continue now.
We now have a stock vsti that's on the money.


3 min test
At start

At 3 min,


Stock VSTi now runs exactly the same as my timer.

Going for lunch.
First round in heavyweight title (polariry test) later in the day.

I've just been following along on this thread [as a potential migrater from SM]. Beside the interesting and important topic ... just wanted to express my respect for your determination [and experience] !

billv wrote:if you get annoyed doing printscreens, I really recommend a small FREE program called
"Gadwin PrintScreen"
http://www.gadwin.com/printscreen/

you can add at yours list of utility for grab the screen this LICEcap from Cockos

Demo how record with LICEcap

Tronic wrote:you can add at yours list of utility

Cheers. Thankyou.

billv wrote:We now have a stock vsti that's on the money.

Well, not quite. in order to be on the money, must be consistant.
Several times this afternoon I have failed to get the exact same result.

So there's still some work to do on this design.

It's still a new thread.
Hopefully, A guru could be working on a better design as we speak

Hello again. Here's an answer from Gol about the apparent drift of waveforms in the playlist.

I believe it's because the display of the clip's content is rounded to ticks. If you switch to 960PPQ it should be better. But really it shouldn't matter, only visual (but I can understand the problem if you used it to check accuracy of something else)

And then

yes it's visual, better trust edison for sample-accuracy

So maybe I do know something.

If the length of the waveforms drawn in the playlist are rounded down ever so slightly then that would create a slight visual drift once the clip gets long enough.

So, in conclusion, better use Edison!

Perfect Human Interface wrote:So maybe I do know something

Always good to suprise yourself

Perfect Human Interface wrote:Here's an answer from Gol about the apparent drift of waveforms in the playlist.

Interesting stuff. can't thank you enough for chasing that one up.
Will try this 960 option and see what the deal is.

Perfect Human Interface wrote:But really it shouldn't matter, only visual

that's great news. I can put my timer back on the table and resume testing
Good for the stock vsti fix too, might be more accurate than i thought.
Dosn't help understanding the Polarity test failures.
Bear in mind the second test came after the stock vsti fix.
So how the hell did i get the same result...?
Both tests I'm only 100% on the first bar...Why?...
Lots of clues still there i have to go through.

billv wrote:then that would create a slight visual drift once the clip gets long enough.

wow...that's some more weight to the comment above "might be more accurate than i thought"

Perfect Human Interface wrote:So, in conclusion, better use Edison!

Will definatly start some comparison testing to see whats going on.

Lots of great news there Mr.Perfect Human Interface.
Thanks again for that.
In light of all this, i will stop building new designs, and get a better understanding
of my test enviroment first.
Cheers mate.
PS:
That might explain too, why all of a sudden after 3 days of "no error" testing,
I suddenly got some drift. the testing method is not consistant!!!!
Wicked...I'm getting excited now.....

Perfect Human Interface wrote:create a slight visual drift once the clip gets long enough.

Confirmed. So......
THIS THREAD IS A FALSE ALARM. Testing method used was in-accurate.

I re-tested the "Stock VSTi"

After 1 minute,
Here's the edison recording sent to playlist, showing drift

Here's the file still in edison, right on the money


So, there is no need for a fix to the VSTi.
DSPR, apologies for implying that your work was sub-standard.
Note: Mr.Perfect Human Interface. Brilliant work mate.
What you have brought to the table here, is easily one of the biggest
revelations for me in the last 6/7 years while working with SM/FS
community.
Please buy yourself a beer or twenty and charge it to my account asap :

Wow, now that's some praise. Too bad I don't drink!

Oh and regarding the first note cancelling, remember those tiny jitters I mentioned due to rounding since tempo and sample rate (and possibly something else??) don't line up perfectly? Well I think it's correct to say that the first instant is the only one where everything is guaranteed to line up 100% and no rounding is involved.

It may end up that the inversion test won't work due to this after all. Then again even with rounding errors I might expect it to make the same errors consistently. But seeing how far you can zoom in in Edison (and knowing that it's accurate), eyeballing it may just end up being the best bet in the end.

By the way, feel free to abbreviate if my forum name is too much work.