Random Morphs shifting and not saving

[soundman516]

I'm having issues with morphs not saving. When I moved to XL v4.2 some of my morphs in a sequences shifted on my pixel tree. I adjust them back to how I had them programmed, I saved and then closed. When I reopen the .fseq or the XML they are back to the incorrect locations. None of the changes stay with the morphs when I save. If I make adjustments to any other effect, those changes do seem to stay. I have opened the sequences in previous versions of XL & the latest XL v4.2.2 but nothing changes.

I opened from my original XML deleted the contents in the timing area, rendered all, reprogrammed the morphs, saved, closed XL and reopened the sequence. It still didn't fix it, even with new morphs being programmed. Yesterday I opened up a backup of the sequence from a previous week and all worked fine, so I saved it as a new file. Today I open it up and my morphs have shifted. Does anyone have any idea of what could be causing this corruption?

Here's a video showing a little of what is happening: https://www.youtube.com/watch?v=M38QSlsD9NA&feature=youtu.be&hd=1

Also, here is a link my RGBeffects.XML and Sequence XML if you care to take a look.
https://www.dropbox.com/sh/q8cvkbp6q022p8c/AACLJbXNK6LRBT61qLj2kIb1a?dl=0

I've also included a screenshot of how the model is setup in the link above.

[RGBNoob]

On your model setup you can start string 4 at channel 451 instead of skipping channels 451-510.

[soundman516]

So I decided to go back to the backup XML file from a couple weeks ago again. I opened this XML and I programmed some in this file as my starting point and all was working fine while just rendering the sequence. I was not hitting save at this point to generate the .fseq. Once I save the sequence and the .fseq is generated, then go back and open from the .fseq, thats when I notice that my morphs have moved. From here on out, when I adjust the morphs and save, they never hold their saved position.

Thoughts?

[Gilrock]

You know now that I got time to watch the video it would have been a lot clearer to me if you had described it as shifting to the left one column.  If you said that I missed it.  I've seen this and I know why it happens but its normally on much bigger models.  The problem is I have to convert from percentages to column numbers and then back to percentages to save the effect and sometimes there are rounding issues.  Where I usually see the problem is on models bigger than 50 columns because the resolution of the value we store is 1% so I can't represent 20.7% for instance.  Once you get to 50 columns each column is only 2% wide so its easy to round off the wrong side of a percentage when doing the converting.  But I'll take a look at your special case and see if there's any way to do the conversion better.

[Gilrock]

Forget everything I just said.  It had me baffled but I tracked it down.  This is kind of an "oh shit" moment.  Everyone's Morphs are going to be screwed up if they opened and saved in any version 4.2.0 or higher.  In version 4.1.10 I changed how morph percentages were handled.  So there is detection code to fix all the percentages for sequences that were older than 4.1.10.  Well the logic to check whether a version is older is wrong.  It got to the 3rd part of the version number and thought the 10 was newer and so it's doing a percentage conversion on everyone's morph values.  Thanks for finding this.....now I'm embarrassed.

[kevinp]

Gil, I fixed a morph effect in one sequence last night to shift one column back into proper location.  Will your fix now affect the changes I just made again?  I'm thinking not, but you just never know with all of the complex calculations going on within the program.

[soundman516]

Forget everything I just said.  It had me baffled but I tracked it down.  This is kind of an "oh shit" moment.  Everyone's Morphs are going to be screwed up if they opened and saved in any version 4.2.0 or higher.  In version 4.1.10 I changed how morph percentages were handled.  So there is detection code to fix all the percentages for sequences that were older than 4.1.10.  Well the logic to check whether a version is older is wrong.  It got to the 3rd part of the version number and thought the 10 was newer and so it's doing a percentage conversion on everyone's morph values.  Thanks for finding this.....now I'm embarrassed.

Gil, no worries! I'm glad you've traced down the problem, and it's better to have found it sooner than later for sure! I'm sure I speak for everyone, these effects are awesome and we can't thank you enough of creating them in XL!

[Steamyrotter]

  Thanks for finding this.....now I'm embarrassed.

The fact you 'owned up' to it, and flagged this up shows you have done the right thing and are open and honest.

That stand you in greater stead, and therefore no embarrassment necessary in my mind..

Keep up the good work