Text scroll for "Credits" shows solid white

[Tim Reid]

Hi folks,

Just programmed a text cue with a (long) list of credits animated to scroll up the screen - but when running the cue it's a just screenful of solid white that scrolls, not the text. 

The text's all there in the inspector but it was glitchy when editing and:

- there's a lot of it - the translation fade cue runs from 0,-16550 to 0, 16550 (I tried over 60 and 30 seconds with the same result)

- it's in a custom font, may or may not be relevant. (I think not - I tried in Arial, same result)

Other factors:

- static titles (ie not scrolling) behave properly, in the same font.

- a much smaller amount of text scrolls properly (0, -2330 to 0, 2330)

So have I just hit some upper limit with such a huge bunch of text?

Thanks for any ideas

Tim

[micpool]

Yes your effective image is too big.

I don't know if this is graphics card dependent but anything above around 15000 pixels high on my machine won't render.

To test just repeat a line of text and copy and paste and test until your screen goes white.

You can see the total pixel dimensions of the text cue in the text inspector

Mic

[luckydave]

As I understand it, the limit here is determined by vRAM, which of course is part of the graphics card. The most common place I've seen this issue is when folks take high megapixel photos straight from a DSLR into QLab, and often on a computer with only Iris graphics, so no dedicated vRAM at all.

[micpool]

Fortunately, you only use vRAM for images that are actually on screen, so there is a simple work around for any size of image

Split your image/text cues into blocks of say, 10000 pixels high.

Put all the blocks in a timeline group and offset their starting positions by -10000 

e.g for a text scroll starting below the screen

Block 1 -5000 (first block  Block 2 -15000, Block 3 -25000. etc.

Then use a relative fade targeting the timeline group with a relative fade

[micpool]

This technique also works well for horizontal scrolling panoramic backgrounds.

I don't know if there is a limit to this but I have got a 120,000x1080 pixel image scrolling perfectly by chopping it  into 23  5000x1080 images, offsetting their starting x translations by 5000 pixels and applying a relative fade to the group containing them of -120,000 pixels. 

An image this size will scroll smoothly from a fast moving 60 seconds to  anything up to  glacial 33 hours at 1 pixel per second

Although I haven't experimented with this, at  the slower speeds it may well be possible  to do, for instance, a 4 layer panorama scroll with parallax by using different speeds for each layer e.g sun, mountains, river, buildings

Mic

[Chet Miller]

FWIW, there is a hard texture limit in most gpus (maybe all still?) of 16,384 pixels in any one direction, so that becomes a limiting reagent in media servers quite often.

[Tim Reid]

Thanks all for the helpful info and suggestions. We ended up cutting down the credits and adding a web link, but I'm now eager for a project to let me try out Mic's cool parallax idea... !

All the best folks

Tim

Tim Reid

he/him/his

t...@timreidvideodesign.com

07941 032 646

www.timreidvideodesign.com

-- 
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
--- 
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/81ede6a8-fff3-4aa8-89ea-4876c4246f75n%40googlegroups.com.

[micpool]

Yes, all my computers have a limit of 16834 pixels in any dimension for an image or the effective size of a text cue.

This is most limiting in horizontal scrolling ticker style text. e.g 250pt text on a 300pixel high strip will only allow about 25 words before displaying as solid white.

By using the multiple cues with offsets technique  I described I'm my previous posts, even on a computer with limited performance e.g a 2018 Mac mini with Intel graphics and 8GB RAM, it is possible to smoothly scroll half a million pixels of text (which is about 20 minutes of text crawl.)

I have been working on a script to automate this, but got stuck with trying to determine image sizes  by using the  image events application,  which seems to have permissions issues in Catalina. (I could probably get it working but it would be a pain for anyone else using the script to have to go through the steps necessary). I have now found a way to use QLab's /cueSize and /surfacesize methods  in OSC queries from AppleScript and parse the dictionary returned, so should have a comprehensive method for automating the programming of large images ( I've got 50,000x2,160 images and 500,000x300 text cues scrolling perfectly) ready in a couple of days.

Mic

[micpool]

Here is a link to a QLab workspace and associated public domain gigapixel images (copyright details in notes of example timeline groups)

 https://we.tl/t-YKtMgR6K2W

The workspace contains :

4 prebuilt examples of large scrolling images in timeline groups

The component cues of these examples for you to apply the script to (Hotkey 1)

Space for you to drag your own images, or to create text cues,  edited for the  maximum width and height  your graphics card allows for textures (I suggest 15000 as an easily remembered value) to apply the script to.

The script cue with a script that :

Allows you to choose the start and end points of the scrolled image, and choose whether these end points are on or off screen

Scales the selected  video or text cues, sized to your screen width or height, depending on direction of travel.

Calculates the offsets for the translations

Puts  the selected cues into a timeline group together with a fade cue to apply a relative fade to all the images to scroll the large image. (The translation in this fade cue determines where the combined image will end, so you can duplicate these cues to move a certain number of pixels in each direction.)

Adds a fade cue if the image ends on screen

Previews the timeline group for you to preview prior to adjusting  fade durations  and curve shapes as required.

The cues are built to the dimensions of the surface the cues are patched to when the script is run. Full screen and lower third surfaces are included which you will need to change in settings for your display and dimensions.

Other things to try include,  assigning a Syphon only surface for the horizontal scrolling text, bringing that surface into a camera cue assigned to the full screen surface  and rotating it 45 degrees, to get a diagonal  single line text crawl. 

You don't need to know anything about scripting to use  the example workspace  but ,for those interested, here's the script: (I couldn't get the OSC queries that get the image and surface sizes to work in shell scripts sending  to port 53535. I don't know if this is because  the text port doesn't process them, or if I just had the wrong escaped characters and quotes etc. As I also needed a scratch pad to deal with the values returned for these queries I just made and deleted network cues as required, as there is no significant time penalty in doing this, and as this is a much clearer and adaptable  method than shell scripting, I think it's a better way of doing things in this instance.)

I have got an image scroll and 2 text scrolls all happening simultaneously, on a very modestly specced Mac mini. If you have a powerful machine  with an eGPU, I'd be interested in knowing how wide an image you can successfully scroll before it gets jerky, and whether you can scroll large panoramas with transparency for parallax scrolling. . My computer will  do a combined  cues size of  50000x2160 images or 500,000x300 text images. smoothly.

--initialise

set oldTIDs to AppleScript's text item delimiters

set theScaledWidths to {}

set theScaledHeights to {}

set cueListFlag to false

set onScreenFlag to false

tell application id "com.figure53.QLab.4" to tell front workspace

set thecues to (selected as list)

--check the selection contains is either all  text , or all video cues, and  is not empty.

if the (count of thecues) < 1 then

display dialog "No text or video Cues Selected!"

return

end if

if the q type of item 1 of thecues is "Text" then

set thecueType to "Text"

else

set thecueType to "Video"

end if

set theparent to parent of item 1 of thecues

repeat with thecue in thecues

if the q type of thecue is not in {"Video", "Text"} or (q type of thecue is not thecueType) then

display dialog "Selection contains cues that are not  all text, or all video cues."

return

end if

if the parent of thecue is not theparent then

display dialog "Selection contains cues that are in a different groups or cue lists "

return

end if

end repeat

if the q type of theparent is "cue list" then set the cueListFlag to true

set theDirection to choose from list {"Left to Right", "Right to Left", "Top to Bottom", "Bottom to Top"} with prompt "IMAGE SCROLL  BUILDER" & return & return & "Check that the " & (count of thecues) & " cues selected are video or text cues, ordered L to R, or top to bottom." & return & return & "Please check carefully that the cues selected are assigned to the same surface (This is not checked automatically!)" & return & return & "Select direction below to make a scrolling  image or text sequence from the selected cues." & return & return & "Choose image start and end points l" default items {"Left to Right"} without multiple selections allowed

if theDirection is {"Left to Right"} then

set theXfactor to 1

set theYfactor to 0

set thereverse to 0

else if theDirection is {"Right to Left"} then

set theXfactor to -1

set theYfactor to 0

set thereverse to 1

else if theDirection is {"Top to Bottom"} then

set theXfactor to 0

set theYfactor to -1

set thereverse to 0

else if theDirection is {"Bottom to Top"} then

set theXfactor to 0

set theYfactor to 1

set thereverse to 1

else

return

end if

set thebutton to button returned of (display dialog "Choose if Start and End points of scroll are on or off screen" buttons {"ON", "OFF", "Cancel"} default button 1)

if thebutton is "Cancel" then return

if thebutton is "ON" then set onScreenFlag to true

if thereverse is 1 then set thecues to reverse of thecues --R to L, or Down motion directions

repeat with thecue in thecues

--store the cue number and temporarily change it for use in the OSC query

set theoldcuenumber to q number of thecue

set the q number of thecue to "tempx"

--pad the line if horizontal text

if q type of thecue is "Text" and theXfactor is not 0 then

set the text of thecue to (text of thecue & "\\")

end if

--to make OSC queries work correctly, create a network cue that will be deleted after use

make type "network"

set theselected to last item of (selected as list)

set the q number of theselected to "ERASEME"

--get the surface size for the surface the cue is patched to

set the custom message of theselected to "/cue/ERASEME/notes \"#/cue/tempx/surfaceSize#\""

set the selected to thecue

start cue "ERASEME"

set AppleScript's text item delimiters to {";", " height = ", "width = "}

set thenotes to notes of cue "ERASEME"

set theScreenWidth to text item 4 of thenotes as string

set theScreenHeight to text item 2 of thenotes as string

--get the image size for the file the cue targets or the size of the text image

set the custom message of theselected to "/cue/ERASEME/notes \"#/cue/tempx/cueSize#\""

set the selected to thecue

start cue "ERASEME"

set thenotes to notes of cue "ERASEME"

set theImagewidth to text item 4 of thenotes as string

set theImageheight to text item 2 of thenotes as string

--set the scaling to fit the images or text on screen.

if theXfactor is not 0 then

set thescale to theScreenHeight / theImageheight

else

set thescale to theScreenWidth / theImagewidth

end if

set thescaledW to theImagewidth * thescale

set thescaledH to theImageheight * thescale

--erase text padding after gap between text cues has been calculated

if thecueType is "Text" and theXfactor is not 0 then

set thetext to text of thecue as string

set thetids to AppleScript's text item delimiters

set AppleScript's text item delimiters to "\\"

set the text of thecue to text item 1 of thetext

set AppleScript's text item delimiters to thetids

end if

--clean up

set q number of thecue to theoldcuenumber

delete cue "ERASEME" of parent of cue "ERASEME"

--set video properties for the cue

set the full screen of thecue to false

set scale x of thecue to thescale

set scale y of thecue to thescale

set end of theScaledWidths to thescaledW

set end of theScaledHeights to thescaledH

end repeat

--initialise accumulators fro running image width and height to enable image offsets to be calculated 

set the widthAccumulator to 0

set the heightAccumulator to 0

--set the translations for each cue

repeat with eachcue from 1 to the count of thecues

set scaledImageWidth to item eachcue of theScaledWidths

set scaledImageHeight to item eachcue of theScaledHeights

set thisCue to item eachcue of thecues

-- if the cues are  text cues then start and end the scroll off screen

if onScreenFlag is true then

set the translation x of thisCue to (widthAccumulator - (theScreenWidth / 2 - scaledImageWidth / 2)) * theXfactor

set the translation y of thisCue to (heightAccumulator - (theScreenHeight / 2 - scaledImageHeight / 2)) * theYfactor

else

set the translation x of thisCue to ((widthAccumulator - (theScreenWidth / 2 - scaledImageWidth / 2)) * theXfactor) + (theScreenWidth * theXfactor)

set the translation y of thisCue to ((heightAccumulator - (theScreenHeight / 2 - scaledImageHeight / 2)) * theYfactor) + (theScreenHeight * theYfactor)

end if

set widthAccumulator to widthAccumulator + scaledImageWidth

set heightAccumulator to heightAccumulator + scaledImageHeight

end repeat

--make a relative fade cue to move all the images together and set its duration to a suitable value to test it

make type "fade"

set thefadecue to last item of (selected as list)

set the do translation of thefadecue to true

if onScreenFlag is true then

set the translation x of thefadecue to -(widthAccumulator - theScreenWidth) * theXfactor

set the translation y of thefadecue to -(heightAccumulator - theScreenHeight) * theYfactor

else

set the translation x of thefadecue to -((widthAccumulator - theScreenWidth) * theXfactor) - (theScreenWidth * theXfactor * 2)

set the translation y of thefadecue to -((heightAccumulator - theScreenHeight) * theYfactor) - (theScreenHeight * theYfactor * 2)

set stop target when done of thefadecue to true

end if

if widthAccumulator is greater than heightAccumulator then

set the duration of thefadecue to widthAccumulator / 500

else

set the duration of thefadecue to heightAccumulator / 500

end if

--make a timeline group ad move  all the cues  all the cues into it

make type "group"

set thegroupcue to last item of (selected as list)

set the autoload of thegroupcue to true

set the mode of thegroupcue to timeline

set the q number of thegroupcue to ""

set the q name of thegroupcue to "Scroll Image Group: " & q list name of item 1 of thecues

set the cue target of thefadecue to thegroupcue

repeat with eachcue in thecues

move cue id (uniqueID of eachcue) of parent of eachcue to end of thegroupcue

end repeat

move cue id (uniqueID of thefadecue) of parent of thefadecue to end of thegroupcue

-- if end of cue leaves image on screen create a fade cue to fade and stop it

if onScreenFlag is true then

set the selected to thegroupcue

make type "fade"

set thefadecue to last item of (selected as list)

set the do opacity of thefadecue to true

set the opacity of thefadecue to 0

set the stop target when done of thefadecue to true

set the cue target of thefadecue to thegroupcue

end if

delay 1

--start the cue to test it 

load thegroupcue

delay 1

start thegroupcue

end tell

--clean up

set AppleScript's text item delimiters to oldTIDs

Mic

[micpool]

For those still interested in long scrolling text cues, here are the final few  bits of the jigsaw

This workspace will emulate a ticker sign on the corner of a building.

It outputs the text to a syphon surface which the camera cue uses as its input to play it behind the led panel png which is just a solid black rectangle with a lot of square alpha holes in it.

To make the text look more like it is filling each pixel of the screen the camera cue has a pixellate video filter applied to square off the edges of the text. In practice, because the text is moving you can set a lower value which makes the text look better close  on screen as opposed to viewed 20 metres away.

That way the  camera pixels and the led grid stay stationary as a lens,  to the  the text moving behind it.

These cues are output to a surface which has a simple perspective screen split to simulate a 90 degree bend at the corner of the building

To replace the sign with text from a text file, press 1 to run this script

The script first splits the text files into cues that are within the pixel width allowed for the texture map for each cue by adding a word at a time to the text cue until the reported text output size width of the cue exceeds 150000, when it will make a new cue and continue.

It then calculates the offsets for these cues and groups them with a fade cue that will apply a relative fade to all cues to move them across the surface.

(Workspace attached. It should play something  on most systems just by changing the display on the triggers surface to your output display, Although the geometry might not match the screenshot above if you use a different size monitor.)

-- Script to convert a text file to multiple QLab text cues within permitted pixel dimensions of a texture.

--set up text format in settings/cue templates/text

--initialise

set oldTIDs to AppleScript's text item delimiters

set AppleScript's text item delimiters to {" ", (ASCII character 10), (ASCII character 8)}

set theScaledWidths to {}

set onScreenFlag to false

--convert text file to cues

set thePixels to 15000

tell application id "com.figure53.QLab.4" to tell front workspace

stop cue "SIGN"

set the selected to cue "SCRIPT"

set theFile to choose file of type "txt" with prompt "Convert a text file to multiple QLab text cues, within permitted pixel dimensions of  texture map. " & return & return & "Select a text file:"

set theFileContents to read theFile

set theCueIndex to 1

make type "text"

set thetextcue to last item of (selected as list)

set the q number of thetextcue to "Temp" & theCueIndex

set the text of thetextcue to ""

set thelinelength to 0

repeat with theword from 1 to count of text items of theFileContents

set the text of thetextcue to text of thetextcue & " " & text item theword of theFileContents

set thelinelength to item 1 of (text output size of thetextcue as list) as integer

if thelinelength is greater than thePixels then

make type "text"

set thetextcue to last item of (selected as list)

set theCueIndex to theCueIndex + 1

set q number of thetextcue to "temp " & theCueIndex

set the text of thetextcue to ""

set linelength to 0

set text of thetextcue to ""

end if

end repeat

end tell

tell application id "com.figure53.QLab.4" to tell front workspace

try

delete cue "TEXT" of parent of cue "TEXT"

end try

set thecues to cues of front cue list whose q number begins with "Temp"

repeat with thecue in thecues

--store the cue number and temporarily change it for use in the OSC query

set theoldcuenumber to q number of thecue

set the q number of thecue to "tempx"

set the text of thecue to (text of thecue & "\\")

--to make OSC queries work correctly, create a network cue that will be deleted after use

make type "network"

set theselected to last item of (selected as list)

set the q number of theselected to "ERASEME"

--get the surface size for the surface the cue is patched to

set the custom message of theselected to "/cue/ERASEME/notes \"#/cue/tempx/surfaceSize#\""

set the selected to thecue

start cue "ERASEME"

set AppleScript's text item delimiters to {";", " height = ", "width = "}

set thenotes to notes of cue "ERASEME"

set theScreenWidth to text item 4 of thenotes as string

set theScreenHeight to text item 2 of thenotes as string

--get the image size for the file the cue targets or the size of the text image

set the custom message of theselected to "/cue/ERASEME/notes \"#/cue/tempx/cueSize#\""

set the selected to thecue

start cue "ERASEME"

set thenotes to notes of cue "ERASEME"

set theImagewidth to text item 4 of thenotes as string

set theImageheight to text item 2 of thenotes as string

--set the scaling to fit the images or text on screen.

set thescale to (theScreenHeight / theImageheight)

set thescaledW to theImagewidth * thescale

set thescaledH to theImageheight * thescale

--erase text padding after gap between text cues has been calculated

set thetext to text of thecue as string

set thetids to AppleScript's text item delimiters

set AppleScript's text item delimiters to "\\"

set the text of thecue to text item 1 of thetext

set AppleScript's text item delimiters to thetids

--clean up

set q number of thecue to theoldcuenumber

delete cue "ERASEME" of parent of cue "ERASEME"

--set video properties for the cue

set the full screen of thecue to false

set scale x of thecue to thescale

set scale y of thecue to thescale

set end of theScaledWidths to thescaledW

end repeat

--initialise accumulators fro running image width and height to enable image offsets to be calculated 

set the widthAccumulator to 0

--set the translations for each cue

repeat with eachcue from 1 to the count of thecues

set scaledImageWidth to item eachcue of theScaledWidths

set thisCue to item eachcue of thecues

set the translation x of thisCue to ((widthAccumulator - (theScreenWidth / 2 - scaledImageWidth / 2))) + (theScreenWidth)

set widthAccumulator to widthAccumulator + scaledImageWidth

end repeat

--make a relative fade cue to move all the images together and set its duration to a suitable value to test it

make type "fade"

set thefadecue to last item of (selected as list)

set the do translation of thefadecue to true

set the q number of thefadecue to "FADE"

set the translation x of thefadecue to -((widthAccumulator - theScreenWidth)) - (theScreenWidth * 2)

set stop target when done of thefadecue to true

set the duration of thefadecue to (translation x of thefadecue) / -250

--make a timeline group ad move  all the cues  all the cues into it

make type "group"

set thegroupcue to last item of (selected as list)

set the autoload of thegroupcue to true

set the mode of thegroupcue to timeline

set the q number of thegroupcue to "TEXT"

set the q name of thegroupcue to "Scroll Large Text Image : "

set the cue target of thefadecue to thegroupcue

repeat with eachcue in thecues

move cue id (uniqueID of eachcue) of parent of eachcue to end of thegroupcue

set q number of eachcue to ""

end repeat

move cue id (uniqueID of thefadecue) of parent of thefadecue to end of thegroupcue

move cue id (uniqueID of cue "TEXT") of parent of cue "TEXT" to beginning of cue "SIGN"

set the pre wait of cue "REPEAT" to (duration of cue "FADE") + 1

delay 1

--start the cue to test it 

load (first cue whose q number is "SIGN")

delay 1

start cue "SIGN"

end tell

--clean up

set AppleScript's text item delimiters to oldTIDs

Mic

[Tim Reid]

Mic, you are as ever a legend and a genius. Amazing what you can do with Qlab. Thanks for the inspiration!

Tim

Tim Reid

he/him/his

www.timreidvideodesign.com

On 19 Mar 2021, at 18:36, micpool <m...@micpool.com> wrote:

For those still interested in long scrolling text cues, here are the final few  bits of the jigsaw

This workspace will emulate a ticker sign on the corner of a building.

<Screenshot 2021-03-19 at 18.22.31.png>

It outputs the text to a syphon surface which the camera cue uses as its input to play it behind the led panel png which is just a solid black rectangle with a lot of square alpha holes in it.

To make the text look more like it is filling each pixel of the screen the camera cue has a pixellate video filter applied to square off the edges of the text. In practice, because the text is moving you can set a lower value which makes the text look better close  on screen as opposed to viewed 20 metres away.

That way the  camera pixels and the led grid stay stationary as a lens,  to the  the text moving behind it.

These cues are output to a surface which has a simple perspective screen split to simulate a 90 degree bend at the corner of the building

-- 
Contact support anytime: sup...@figure53.com
Follow QLab on Twitter: https://twitter.com/QLabApp
User Group Code of Conduct: https://qlab.app/code-of-conduct/
--- 
You received this message because you are subscribed to the Google Groups "QLab" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qlab+uns...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/qlab/193e7473-ab95-4f94-99e8-3775513d6164n%40googlegroups.com.
<Screenshot 2021-03-19 at 18.09.29.png><Tickers.zip><Screenshot 2021-03-19 at 18.22.31.png>