Sean Meighan
Welcome => Do You Need Help? Post it here => Topic started by: Ebuechner on April 08, 2016, 11:45:24 AM
-
While sequencing I've attempted to use the Shimmer and twinkle and I've noticed that they really don't look very good on the preview or even the lights that I have hooked up. I have a sequence that would Shimmer just fine using light O Rama software then I converted over an Xlights does not represent it very well. It almost seems like a refresh rate problem in Xlights. So I'm asking am I missing a setting or is there any suggestions? And does it act differently if it's played from the schedule? Until now I've done my sequencing between X lights and light O Rama and scheduled my show through light O Rama. My channel count is up over 12,000 now.
-
I'm curious if you're even using the effect settings box to adjust these.
The output of the effect depends on the length of the effect and settings like colors chosen, duty factor, cycle count, number of lights, steps, etc etc.
The beauty is that you can fine tune the effects to your own desires.
-
While sequencing I've attempted to use the Shimmer and twinkle and I've noticed that they really don't look very good on the preview or even the lights that I have hooked up. I have a sequence that would Shimmer just fine using light O Rama software then I converted over an Xlights does not represent it very well. It almost seems like a refresh rate problem in Xlights. So I'm asking am I missing a setting or is there any suggestions? And does it act differently if it's played from the schedule? Until now I've done my sequencing between X lights and light O Rama and scheduled my show through light O Rama. My channel count is up over 12,000 now.
Are you using a 25ms seq or 50ms?
Can you try something: can you create a view with either no elements or a single, relatively unused element. Select that view, then hit play and see if the House Preview seems to look better. I've been playing around a bit and on my 4 year old machine (so relatively decent), playing the "Star Wars Funk" sequence at 25ms with my full "All Elements" view is dropping every other frame. That would certainly make the shimmers not work since they are every other frame.
Doing some timing, it's taking about 2-3ms to update BOTH preview windows, but it's taking another 32-40ms to update all of the stuff in the grid. That includes the "Time" in the upper left, the play markers on the timeline, wave form, and grid. If I use a view with only a single model, it takes about 12-14ms. 7ms of that is just updating the "Time" in the corner.
Thus, we have some stuff to think about:
1) Is the "Time" in the corner useful? If so, should we update it less often? If not, eliminate it?
2) How we draw the effect grid needs a ton of work. I've already been chatting to Keith about it, but this likely is enough to convince me it's more necessary. We really need to aggregate all the "lines" together, all the rects, etc... so things can be sent to the graphics card in large batches. The reason the House/Model previews are so quick is we DO aggregate things there. My 1600 pixel mega tree is literally rendered with just 7 opengl calls. Each EFFECT displayed on the grid is currently about 35 opengl calls PLUS the effect background if one is displayed.
-
First off I'm using a relatively new computer that's not built to minimum standards. 32 gigs of RAM with a newer AMD A10 FM 2+ processor 7860k. I'm running dual Samsung monitors at 1080. My mega tree is 36 strands with 56 pixels each I have a sequence with very few effects on it that I imported from Lor. With the light O Rama software it ran fine and rendered fine in their visualizer and output it to the lights just fine. I imported into the X lights and I can't get the Shimmer to look very good at all or output to the lights properly. (I imported at 50ms) I'm still learning the X Lite software but I'm not completely green. I played with a new sequence trying to get a good Shimmer in xlights without much luck I was adjusting the colors duty cycle timing andeverything
-
2 seasons ago when I became predominately RGB, I stopped using shimmer and removed it from all sequences.
I came from an LOR protocol start with AC controllers. Shimmer and Twinkle were fine because they were designed for incandescent bulbs and how they cycle.
Shimmer on LEDs and Pixels isn't that great, even in a pure LOR setup.
Twinkle, on the other hand, seems to still work well on a string of pixels. (just not like it did with LOR protocol)
-
I guess I'm just wondering why a sequence in Lor software will render and output to the lights just fine. The same sequence on the same computer with the same network setups doesn't work well in xlights? I'm trying to compare apples to apples. I'm hoping that it's just a setup issue that I haven't found yet.
-
LOR does Twinkle and Shimmer in hardware....xLights does not use hardware specific features. I say talking about this is pretty useless without us seeing your files. Upload the xml file for the sequence you imported and the xlights_rgbeffects.xml file so we can see what you really have there.
-
Before anybody asks I just toggled all the way down to the nodes and checked the Shimmer effect it's at a 50% duty cycle timed perfectly at 50ms so it's all channels on at 50ms then all channels off at 50ms and repeat.
-
here are the files
-
Ok so it looks like you are importing as a Data Layer. That means I won't see any effects or data in the sequence you uploaded and the ISEQ file is usually a big data file.
So are you saying you see the on/off shimmers in the node data?
And are you trying to playback the data from xLights over a LOR network? I haven't dug into that code but from all the reports I don't think we run as well over the RS-485 network as LOR does. Many people have lag issues. Most of us run from an FPP so we use E1.31 network data which is awesome. I'm sure it could be improved but it's hard to find a developer willing to sink time into something they don't use.
-
All this started because I'm going away from the Lor Network everything is on E1:31. I've even abandon my LOR ac controllers because they didn't work well in DMX mode. I just built some Renard plus 32 controllers and programmed them with DMX protocol. I was thinking about going straight X lights butt I need to address this one last performance issue before I pull the plug on LOR.
-
How big is the LMS file? If it's not huge upload it and I'll import it so I can see the data.
-
Even if I take everything out but the mega tree it's still too big of a file
-
https://drive.google.com/open?id=0BwDgEpqsgNhfY2Z3V0xFb2MwUUU
-
I'm not sure what Gil is expecting to see from the LMS file. A shimmer in LMS will be the 50% on/off per frame thing that you are seeing with xLights. That's exactly how LOR outputs shimmer to DMX devices that don't have the hardware shimmer (someone verified that and sent the screen shots and similar to show it). Thus, I'm sure that what we're doing is the same as LOR for that. That said, if we have to drop frames (during preview or playback), then shimmer is going to look like crap. I'd suggest (as a test) taking the fseq and running from FPP to real lights to see how it looks that way. Alternatively, create a playlist and play it from there with output tuned on in xLights. That would avoid all the video rendering and just blast out the data.
Twinkle is different as there is some randomness involved in that.
-
So I took your suggestion and I played it directly to the lights from the player it seemed to look a little better. Could this be partially a network problem? I'm running a dedicated gigabit Ethernet card on the computer on a static IP out to a Netgear gigabit switch then out to my controllers I have everything running on a static IP.
-
I'm not sure what Gil is expecting to see from the LMS file.
I wanted to see what the data looks like instead of guessing from a description where for several posts I couldn't tell if it was imported as effects or what. But I'll quit looking at it since you and Keith are discussing how to fix it.
-
I've continued to test back and forth with different combinations with the same sequence between Lor and Xlights. I hate to say it but Lor is outperforming X lights when anything has to move fast or Shimmer. Is Xlights not refreshing the network as much as light O Rama ? I even checked with my 6 Port Bridge on packets received. There was significantly less packets received using xlights versus Lor. I am not picking on Xlights at all in fact I want to move over to using it exclusively but I would like to be able to bump up its performance or know that there's a fix coming down the road.
-
I've continued to test back and forth with different combinations with the same sequence between Lor and Xlights. I hate to say it but Lor is outperforming X lights when anything has to move fast or Shimmer. Is Xlights not refreshing the network as much as light O Rama ? I even checked with my 6 Port Bridge on packets received. There was significantly less packets received using xlights versus Lor. I am not picking on Xlights at all in fact I want to move over to using it exclusively but I would like to be able to bump up its performance or know that there's a fix coming down the road.
When you create the sequence, select 25ms instead of 50ms. That will double your frames which could smooth things out for some of the fast moving stuff. If you edit the raw XML, it's also easy to change it to 10ms if you really want, but I don't recommend it.
LOR can go down to 10ms in SOME cases. The problem is that a full DMX universe takes 21ms to send out so once you have any DMX things, going below 25ms makes no sense at all. For the most part, we've always defaulted to 50ms as some of the older pixelnet stuff would only work at that rate plus some of the other technologies that were used (renard with Xbee's for example) would need significantly reduced channel counts. 50ms was always a decent compromise. Most "normal folks watching the show" could barely tell the difference.
-
You're just saying that I'm way too picky ;)
-
All the developers of xLights playback our sequences using the FPP because it kicks the butts of both LOR and xLights playback.
-
For us newbies out there, can you explain how the FPP "kicks but" over the other players?
I'm aware that running a show off the Pi means that you don't have to keep a dedicated computer running the show. But what are the other benefits?
-
Running a show from LOR or xlights you have the Windows operating system underneath the code. When you run the show from a FPP it is running Linux. Linux is a better OS for being deterministic (there isnt as much strange "What is my computer doing" stuff). The FPP has been tested to being able to output 250K channels with no lag. This on a $30 computer. Here is a table showing the current SBC (Single Board Computers) that can run the FPP software http://nutcracker123.com/nutcracker/rgb_controllers/fpp_comparison.html
I started playing my show on the FPP back in 2013 because
1) It works
2) It does not tie up a computer for 8 weeks
3) You can control it with your phone
4) You can have hundreds of different kinds of sensors plug directly into it
Try playing a 100,000 channel show on xlights,LOR and FPP you will see the difference.,
What if your show is only 200 channels? Then use any of the above three and it works fine.
-
Thanks Sean--so one of the big differences when running a "large" (100K channels+) show, is that there can be lag in Xlights/LOR that you won't have with FPP?
-
Thanks Sean--so one of the big differences when running a "large" (100K channels+) show, is that there can be lag in Xlights/LOR that you won't have with FPP?
That is one, but there are so many more reasons to use the FPP computers.
1) There are not that many plug in interfaces for windows boxes, the boards you get are expensive. The raspberry pi and beagle bone were targeted to teh DIY community. Just look at some of these and notice how cheap.
https://www.adafruit.com/categories/35
They can ALL be integrated into the FPP., You want a light beam to launch a sensor, a pressure matt, a ultrasonic sensor, all easy and supporte don FPP, not easy on winodws.
2) You can control the FPP using a phone or ipad
3) Each fpp can drive a 1080p projector. The movie playing on this projector can be synced to +/- 1 frame to your running sequence.
4) You can have 1 master fpp and thousands of slaves. Each fpp can drive 250K channels and a projector. So how about a show of 250 million channels and 1000 projectors? This is supported today.
I run 1 master and 2 slaves (2 projectors: Virtual santa and garage door).
5) The FPP software on a beagle bone black can drive 64 panels (16x32) of P10. That is 32,768 nodes or 98,304 channels. No other software in our industry , that I am aware, can run 98K channels of P10.
6) Your show computer can be run on a small battery. Get the $5 raspberry zero and use a battery for a cell phone and you can run your show player for over 9 hours. You could have this on a person in a marching band, or dancers on a stage.every dancer would be a FPP slave maybe running from the master off stage.
So you see the FPP is special not just that it can run a show without lag, it has huge possibilities. I look forward to see what people have done this year. Thanks to Dave Pitts for inventing the FPP in 2013 and to Capt Murdoch and his team for improving and enhancing it these last few years. Dave handed off his FPP project to the capt in the fall of 2013. Dave no longer has anything to do expect to be a happy user.
-
All good points that I hadn't realized yet.
I haven't played with FPP yet, as I would presume that I want the Rasp Pi 3, and it doesn't appear to be supported just yet.
Am I okay holding off purchasing a Rasp Pi 3 (I think it's still early in the season)?
-
No reason to wait for a Pi 3. For the average show any of the older Pi's work fine.
-
as gil says for 99% of the people running shows
Model B
Model B+
Model 2
Model Zero
All run over 100,000 channels.
At $30, just start getting a collection. I have 6 now. Last year I ran my 45K channels from a Model B+
Eventually ill get a model 3, but why do i need the extra power? If just playing a movie or lights, all work.
If you want to type a persons name and have it overlay-ed the current running show or if you want a virtual matrix you could use the extra horse power.
-
But if I'm going to buy one new, shouldn't I get the most current?
My understanding is that the P3 has built-in WiFi. That sounds like something I would want, no?
-
yep, you will want the pi-3 eventually, just a couple of caveats
1) If you want to get started and learn about 95% of all the FPP stuff (how to configure, how to run, setting up playlists, schedule them , .etc.) you can started right now on the other models.
Will the pi-3 work right now, probably if you tweak. Learning now i think you want the stuff that is rock solid so you dont chase phantoms.
2) New releases for pi-3 can have bugs that you have to work through.
3) I dont use wifi on any of my FPP's. One cable to the switch in the garage and they are all on line. People do setup and run wifi, it was just one more thing to worry about , I chose to skip it. This means wifi is not that important to me
If you dont mind working through some of the early life cycle stuff, get a p-3
-
I'm going off the assumption that it's "still early enough in the season" to wait for the full P3 support to come out, and time for all the kinks to be ironed out.
Come September-ish, I would assume that it would be time to make a purchase, and start getting to know the player. Or do you think that's too late for a semi-technical noob?
-
You can't say that the Pi3 is not supported as I have 2 of them running FPP using the developer branch and v1.8 was just released which officially has support for the 3 in it. I am using mine to test P10 panels using the Pi matrix adapter from Ron's Holiday Lights.
I was Micro Center a few days ago. I bought another Pi3 for $29.99 (only available for instore purchase). A bunch of Pi2s were sitting there in a bin and marked at $34.99. Considering that I would have to buy a wifi adapter for at lea st $10, the Pi 3 was $15 cheaper than the Pi2.
-
I'm just going off what's posted here:
http://nutcracker123.com/nutcracker/rgb_controllers/fpp_comparison.html
Is this not up-to-date?
-
You guys just sent me off in a whole new Direction I guess I'm going to have to learn something new
-
You guys just sent me off in a whole new Direction I guess I'm going to have to learn something new
If you truly want to leave lor, then yes, you are going to have to learn something new.
It makes it really hard when you still have the lor mindset. I jumped those same hurdles. I decided to quit lor like a bad habit. Best decision I made.
James
Sent from my SM-G900V using Tapatalk
-
I just took a sequence and imported it into xlights at 25ms that made a huge difference. I'm thinking that was my problem while trying to create a shimmer effect within Xlights I was starting off at 50ms. Time to get rid of the LOR mindset. If I go in the FPP direction which one of the many units out there is a good starting point to learn from possibly with the most support. B B+ 2 or other?
-
I just took a sequence and imported it into xlights at 25ms that made a huge difference. I'm thinking that was my problem while trying to create a shimmer effect within Xlights I was starting off at 50ms. Time to get rid of the LOR mindset. If I go in the FPP direction which one of the many units out there is a good starting point to learn from possibly with the most support. B B+ 2 or other?
You will get the most help over at falconchristmas.com. That's where all of the fpp guys hang out. I believe the pi3 is supported now. I got mine, b+, at Amazon. It's really up to you. I would get the latest, up to date stuff.
James
Sent from my SM-G900V using Tapatalk
-
I have a Pi Model B, B+, 2, and 3. I ran my show this last year from the B+. I like the case it's in and haven't gotten a good case for the newer models. I see no reason to wait for a 3. My Pi2 and Pi3 are still sitting in the boxes. You can get a $10 wifi dongle from Adafruit that doesn't even stick out more than a 1/4". You definitely don't really need the extra horsepower.
-
I just ordered a pi3 off Amazon $65 got me the board case power supply heatsink and 32 gigs of memory. I've been thinking about playing with one of these anyhow. For that price that's a lot of entertainment or frustration probably both combined into one
-
I'm just going off what's posted here:
http://nutcracker123.com/nutcracker/rgb_controllers/fpp_comparison.html
Is this not up-to-date?
For the latest pi info you should always go to falconchristmas.com
-
I just ordered a pi3 off Amazon $65 got me the board case power supply heatsink and 32 gigs of memory. I've been thinking about playing with one of these anyhow. For that price that's a lot of entertainment or frustration probably both combined into one
Some of the SD cards that come with a bundle aren't any good. You want to get a Class 10 SD card so check it when you get it. Also need to get a good USB flash stick. I get them at Walmart and they have a PNY one I really like because it only takes up the height of the USB port itself and is in a silver metal case.
-
Well I think the light bulb finally clicked on. I took a previously made up sequence in Xlights opened the XML file modified it from 50ms to 25ms opened it up and hit render all then played it. (I noticed a huge difference) I'd like to thank everybody for helping me out with this and I think it's time I start using the software made for the grown-ups
-
I just ordered a pi3 off Amazon $65 got me the board case power supply heatsink and 32 gigs of memory. I've been thinking about playing with one of these anyhow. For that price that's a lot of entertainment or frustration probably both combined into one
Some of the SD cards that come with a bundle aren't any good. You want to get a Class 10 SD card so check it when you get it. Also need to get a good USB flash stick. I get them at Walmart and they have a PNY one I really like because it only takes up the height of the USB port itself and is in a silver metal case.
I ordered the SD card separate I'm pretty sure I got a class 10. I'm sure I'm going to have more questions but thank you for the advice
-
I think it's time I start using the software made for the grown-ups
That a boy! ;D
Sent from my SM-G900V using Tapatalk
-
Well I think the light bulb finally clicked on. I took a previously made up sequence in Xlights opened the XML file modified it from 50ms to 25ms opened it up and hit render all then played it. (I noticed a huge difference) I'd like to thank everybody for helping me out with this and I think it's time I start using the software made for the grown-ups
Yeah I'm not sure about that one. I don't believe changing an existing sequence from 50ms to 25ms is going to change the resolution of the existing effects. They would still all be on their 50ms boundaries.
-
Yeah I'm not sure about that one. I don't believe changing an existing sequence from 50ms to 25ms is going to change the resolution of the existing effects. They would still all be on their 50ms boundaries.
I don't know what to tell you it sure looked like it made a difference to me. Also when I tried the Shimmer I was finally able to get a good output that look decent. But somebody broke the text again. Wouldn't rerendering all the effects change them to the 25ms?
-
It won't change where an effect starts or stops. Just doing a quick look at the Shimmer code it doesn't appear that it would change behavior based on the minimum timing division. You have to change the duty cycle parameter to change it's behavior.
-
I had issues with my p10 panels when playing from xlights/bridge mode. It was fine when playing from fpp/pi. I changed to 25ms and played from xlights and it worked fine. I didn't change the effects, or resolution. So it's obviously doing something different.
James
Sent from my SM-G900V using Tapatalk
-
if I set a sequence at 50ms and I try to do a shimmer it just looks like a random flashing. But if I set it at 25ms it's not perfect but it's close. It's almost like there's two internal clock speeds that aren't working within multiples of each other at 50ms. I'm not making this up it's made a difference. At 25ms it seems like there's more hits than misses.
-
Well feel free to expand the models to the node level and inspect the data and you could screenshot it and show us how the effect is rendered differently. I got other things to do than investigate this.
-
https://youtu.be/CARWeTM9Bos. This video might explain a little bit the first part of the video is a sequence at 50ms. The second part of the video is the exact same sequence modified to 25ms. Both parts of the video are running the exact same effect in the exact same way. Because of the frame rate on the video it'll be a little tough to tell but there's no reason for anybody to get upset with me about it there is very clearly a difference. There was also a very clear difference in the way the effects rendered when I played the sequence.
-
https://youtu.be/CARWeTM9Bos. This video might explain a little bit the first part of the video is a sequence at 50ms. The second part of the video is the exact same sequence modified to 25ms. Both parts of the video are running the exact same effect in the exact same way. Because of the frame rate on the video it'll be a little tough to tell but there's no reason for anybody to get upset with me about it there is very clearly a difference. There was also a very clear difference in the way the effects rendered when I played the sequence.
Nobody is upset with you. If you read enough of Gil's posts you will see that's how he is. Don't take it personally.
James
Sent from my SM-G900V using Tapatalk
-
At this point I'm happy that I figured out how to make it work properly.
-
Yeah I'm not upset I was just trying to save a lot of back and forth talking and guessing. If you show me the data on the grid that tells me more than 1000 words and would be even better than that video. For one I would see whether you are running imported data layer data or whether it's the xLights shimmer effect. And you know that rings a bell. If you imported a data layer and then switched the sequence timing then it squishes all the data in half so yes it will play twice as fast but halfway through the sequence you won't have any imported data because it played back double speed.
-
I had issues with my p10 panels when playing from xlights/bridge mode. It was fine when playing from fpp/pi. I changed to 25ms and played from xlights and it worked fine. I didn't change the effects, or resolution. So it's obviously doing something different.
Did you change FPP's bridge mode timing to 25ms? FPP uses a fixed channel output timing when run in bridge mode. By default this is 50ms but you can change it on the advanced settings page.
-
I had issues with my p10 panels when playing from xlights/bridge mode. It was fine when playing from fpp/pi. I changed to 25ms and played from xlights and it worked fine. I didn't change the effects, or resolution. So it's obviously doing something different.
Did you change FPP's bridge mode timing to 25ms? FPP uses a fixed channel output timing when run in bridge mode. By default this is 50ms but you can change it on the advanced settings page.
Yes, but it made no difference. Matter of fact, it wouldn't stay set. I would change the setting, save, then leave the page and come back and it would go back to 50ms.
James
Sent from my SM-G900V using Tapatalk
-
Yes, but it made no difference. Matter of fact, it wouldn't stay set. I would change the setting, save, then leave the page and come back and it would go back to 50ms.
I made a note to check this out, it may just be a UI issue.