Beat
Beat
Posts: 165
|
I use "Next Variation" quite a lot. It is a super great feature (Topaz had it also, but stupidly dropped it in the Studio versions of it's filter).
Ever better in FF is the possibility to choose between Randomize levels : Strong, Medium and Little.
Now I experience a snag (bug?) :
I apply "Next Variation" in the "Strong randomize" mode. As soon as I get something I like, I switch to "Little randomize" mode. Now I'd expect the next variation to be quite close to what I had. Not so, the "Strong randomize" mode is applied once more, and then only with the next variation, the "Little randomize" mode is applied. But that is a close variant on the Strong Variation, that has taken me away from what I liked, so no use to stay close that that version then.
This defeats the purpose for me of having different Randomize levels.
I think the bug is that the Little Randomization is not effective as soon as it is selected, only after one more Randomize step. For me, this is a very big problem, as I cannot stay close to a variation that I have found and like.
I attach 3 images to show my case :
Random 1 is a variation I got with the "Strong randomize" mode. (shown in this post)
Then I switched to the "Little randomize" mode in the popUpMenu.
Random 2 (Low) is then taken with the "Little randomize" mode active. (shown in next post)Note that this image is completely different from Random 1.
Then I pressed the button "Next variation" again (still in "Little randomize" mode) and now FF does what it should do, making a variation (Random3 (low)) that is close to the previous one. But it is not close to Random1, which is what I want and need.
TheBeat  The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: November 30, 2019 12:30 pm |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
Here is img Random2 (low)  The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: November 30, 2019 12:30 pm |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
and here is Random3 (low)  The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: November 30, 2019 12:31 pm |
Details
E-Mail
|
emme
|
Yes, I have noticed this too. Seems to be a problem with IntSliders specifically. Or perhaps it's just more noticeable with fewer steps. Maybe the internal values are getting incorrectly rounded down somewhere? The error seems to always decrease the value, never increase. The same thing also happens after the user changes the values manually and clicks next variant.
This should probably be moved to Bugs and Problems.
|
Posted: November 30, 2019 1:34 pm |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
Thanks emme. I will have to look into a bit more. I might have found more things, maybe also a way to go around the problem. Plus there is something with the (pseudo) random generator. I will get back on this important subject when I have done some more investigation (thanks for the IntSlider tip : I will look into that as well. Also Switches).
As soon as I know more and I still judge that is is a bug, I will switch to the BUGS & PROBLEMS part. The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 2, 2019 3:53 pm |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
emme, I did a little more experimenting :
Switch with IntSlider :
Changes only with High Randomization if the switch has only 2 inputs.
If more inputs : the jumps also occur in Medium and Low Randomization, the distance of the jumps relate in some way to the Max of the IntSlider (switch).
BUT, and there I think is the error behavior that started this thread and pointed out by you in your post :
•• If a IntSlider (switch) is in a higher position than 1 in High Randomization mode, and the user switches to Medium- or Low Randomization, the next time the "Next variant" button is pressed, the switch is set to a lower input. This of course changes everything, as switches often represent very different paths. After this initial setting to a lower input, the switch varies a little (depending on the max used for the IntSlider) in Medium mode, but stays more or less put in Low mode.
I tried with a Slider control, but the same behavior occurs.
My proposal would be to have a choice in each controlling component as to which level of Randomization it can react. So if I would set a IntSlider to 'Medium randomization', this would mean it cannot budge in Low-, slightly in Medium- and strongly in Strong randomization. This way we could build filters that respond well to the next variant button.
There needs to be some predictability in the unpredictability, don't you think
Will be continued.
TheBeat The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 8, 2019 10:38 am |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
I will also put the 'error' in the bugs & problems, but first I want to get to the bottom of this. The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 8, 2019 10:39 am |
Details
E-Mail
|
emme
|
Yup, I'm sure this is a bug and not intended behavior. It kinda ruins any fine control with the randomizer. I usually work around this by locking IntSliders when randomizing.
Are you sure the problem occurs with Sliders too? For me Sliders seem to work fine, but I might be wrong.
I think fixing this behavior would be enough. No need for extra features once the randomizer works correctly.
I'll post about this in the bugs and problems.
|
Posted: December 8, 2019 1:44 pm |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
Quote |
---|
emme wrote : Yup, I'm sure this is a bug and not intended behavior. It kinda ruins any fine control with the randomizer. I usually work around this by locking IntSliders when randomizing. |
I agree and locking seems a good idea in these circumstances.
Quote |
---|
Are you sure the problem occurs with Sliders too? For me Sliders seem to work fine, but I might be wrong. |
I find partly strange behavior with Sliders, as well as with IntSliders. They do not do exactly the same (of course also due to the randomization process, so only avergaes can be observed). My guess so far would be that the IntSlider moves position is about as much as the Slider, also in first move after switching to a lower Randomization.
I have only tested the two sliders controlling the input of 2 switches.
Another thing is the Pseudo Random generator. It seems that in advance a list of Randomized numbers is made. At least, if I do let's say 5 new variations (A, B, C ,D, E) and the step back, of course I get these in reversed order. But when I press New Variation again, I get again A, B, C ,D, E in exactly the same order and of course the same settings. I think that once you go forward ('New Variation'), you should get a New Variation
emme, yes please post this in Bugs & problems. Great. The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 8, 2019 2:24 pm |
Details
E-Mail
|
GMM
Moderator
Filter Forge, Inc
Posts: 3491
|
Quote |
---|
Beat wrote:
It seems that in advance a list of Randomized numbers is made. At least, if I do let's say 5 new variations (A, B, C ,D, E) and the step back, of course I get these in reversed order. But when I press New Variation again, I get again A, B, C ,D, E in exactly the same order and of course the same settings. |
That's correct, and this behavior is intended. Variations are calculated based on a random seed which is invoked once after a filter is selected. To get another seed and a different sequence of variations, just switch to another filter and back.
|
Posted: December 9, 2019 5:11 am |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
Thanks GMM for clearing this up. Switching to another filter and back is not a problem for me. It's good to know when the random seed is invoked. The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 9, 2019 8:25 am |
Details
E-Mail
|
emme
|
Beat, what version of FF are you using? I just installed FF9 and looks like this is fixed. I'll do some more testing later, but seems like the randomizer is working as expected now.
|
Posted: December 9, 2019 8:48 am |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
I have also upgraded to version 9 (took the life time upgrades, as FF is just an incredible tool for me) and did the testing in version 9. I will also look again.
I appreciate very much that the moderator is keeping in close touch with the community and in my experience always jumps in with technical info if needed  The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 9, 2019 9:25 am |
Details
E-Mail
|
GMM
Moderator
Filter Forge, Inc
Posts: 3491
|
Thanks Beat, appreciate your feedback
|
Posted: December 9, 2019 9:52 am |
Details
E-Mail
|
Beat
Beat
Posts: 165
|
Quote |
---|
emme wrote : Beat, what version of FF are you using? I just installed FF9 and looks like this is fixed. I'll do some more testing later, but seems like the randomizer is working as expected now. |
While testing I also found that the behavior of the randomizer is much better in FF 9. Many filters behave as expected now, but not all. With some filters I still get the old and odd behavior. I will see if I can pinpoint a specific component that might cause the problem sometimes. The Stone Age did not end because they ran out of stones. Niels Bohr
|
Posted: December 16, 2019 4:59 pm |
Details
E-Mail
|