998 replies to this topic

[IanSav]

Hi Foxbob,

 

And on my receiver et11000 4K there is this problem.The cursor in the keyboard moves with a delay.To solve this problem I replaced MultiContentEntryPixmapAlphaBlend on the MultiContentEntryPixmapAlphaTest.And the cursor began to move quickly.

 

Did you make these changes to the VirtualKeyBoard code?  Does the code change affect the overlapped highlighting?

 

Regards,

Ian.

[IanSav]

Hi All,

 

It is looking like a lower level code or hardware issue is driving this bug.  Do people here think this something that can be fixed in the hardware driver?  It will be very frustrating if the GUI code needs to be different for different hardware! 

 

Regards,

Ian.

[WanWizard]

Ian,

 

I can't recall seeing these issues in the beginning, I do a lot of renames of recordings, but the VU+ Duo 4K is my main box, and there it isn't obvious.

 

If I compare my SAB Alpha with my VU+ Duo 4K (running exactly the same image/enigma), there is a substantial difference in response time.

[IanSav]

Hi WanWizard,

 

This sounds like what Foxbob reported.  May I change a copy of VirtualKeyBoard and give you the .py code to test?  If you see no visual degradation but experience a speed improvement I will submit the code change to all images.

 

Regards,

Ian.

[WanWizard]

Sure.

[IanSav]

Hi WanWizard,

 

Please see if this change helps.

 

@Foxbob, does this change match what you changed?

 

Regards,

Ian.

 

Attached Files

•  VirtualKeyBoard.py   59.54KB   2 downloads

[foxbob]

Yes, I did it at home.But some other receivers do not have this problem.

[IanSav]

Hi Foxbob,

 

Yes, I did it at home.But some other receivers do not have this problem.

 

This seems like a "harmless" change.  I ran a quick test and could not see a visual difference.  Did you notice anything look different?

 

If this change helps or fixes the problems Ims and WanWizard reported I will commit this change to all images.

 

Regards,

Ian.

[foxbob]

I seem to have no problems,but maybe someone will change something.Find the reason for this.There may be a problem in the drivers.Because I do the same on some skins-replace Blend to Test.Because it is impossible to move the cursor in the menu.

[ims]

 

@ims try change in skin alphatest="blend" to alphatest="on" or remove the parameter alphatest: https://github.com/O.../skin.xml#L1616

I have one user complained that with blend there are problems in VK on slower receivers not just on hisilicon.

 

It has no effect, it is small png with "Text", it is not VK element.

 

 

@Ian ... change MultiContentEntryPixmapAlphaBlend to MultiContentEntryPixmap for all 3 self.bg_xxx ... it is first acceleration

[WanWizard]

Hi WanWizard,

 

Please see if this change helps.

 

Quite a bit faster on the SAB. Now trying the SF8008.
 

edit: SF8008 is a bit faster too, but the flashing is still there...

Edited by WanWizard, 1 June 2020 - 15:07.

[IanSav]

Hi Ims,

 

 

@Ian ... change MultiContentEntryPixmapAlphaBlend to MultiContentEntryPixmap for all 3 self.bg_xxx ... it is first acceleration

 

The code above already has "blend" changed to "test".  I think you are suggesting to remove all blending from the base image for the VKB sell backgrounds.  But don't they have to be blended with the window background?

 

Regards,

Ian.

[IanSav]

Hi,

 

Given that this issue appears to be related to a hardware or possibly a driver issue, is this something that OpenPLi administrators can refer to the manufacturers for resolution?

 

Regards,

Ian.

[WanWizard]

Might also be hardware, i.e. refresh rate of the framebuffer?

 

If I try to make a video with my phone, the flashing isn't really visible on the video, but it is when you look at it.

[ims]

 

Hi Ims,

 

 

@Ian ... change MultiContentEntryPixmapAlphaBlend to MultiContentEntryPixmap for all 3 self.bg_xxx ... it is first acceleration

 

The code above already has "blend" changed to "test".  I think you are suggesting to remove all blending from the base image for the VKB sell backgrounds.  But don't they have to be blended with the window background?

 

Regards,

Ian.

I think, there is not needed it for key backgroud and window background, imho

M.

[IanSav]

Hi WanWizard,

 

So the flashing issue is still there with the modified code?

 

Has OpenPLi looked at the Enigma2 build options that OpenViX use to get around the hardware / driver issues on HiSi boxes?

 

Regards,

Ian.

[IanSav]

Hi Ims,

 

I think, there is not needed it for key backgroud and window background, imho

M.

 

I will create another version of VirtualKeyBoard and post it here for more testing and review.  I won't commit any changes until there is some good testing of this new alternative on as many boxes as you all have available for testing.  (I only have a Zgemma H7C for testing, and it works fine.)

 

Regards,

Ian.

[ims]

Hi Ims,

 

I think, there is not needed it for key backgroud and window background, imho

M.

 

I will create another version of VirtualKeyBoard and post it here for more testing and review.  I won't commit any changes until there is some good testing of this new alternative on as many boxes as you all have available for testing.  (I only have a Zgemma H7C for testing, and it works fine.)

 

Regards,

Ian.

Yes. On H7 no problem... I found it on hisil boxes.... Hisil Zgemma H9 was almost OK, but GigablueTrio you could saw on video...

M.

[IanSav]

Hi Ims,

 

Is this what you meant?

 

I can't see any issues with a very fast test after changing the code.  Still it would be good for this to be tested on a variety of boxes with a variety of skins.

 

(I suspect that this code will not be faster or better than the other version.)

 

Regards,

Ian.

Attached Files

•  VirtualKeyBoard.py   59.54KB   1 downloads

Edited by IanSav, 1 June 2020 - 15:22.

[IanSav]

Hi,

 

Can anyone explain / define the difference between "AlphaTest" and "AlphaBlend"?  How do they differ?  Why would you use one over the other?

 

I haven't noted any visual differences between using either.

 

Regards,

Ian.

Edited by IanSav, 1 June 2020 - 15:33.