Jump to content

livb89

Resident
  • Posts

    79
  • Joined

  • Last visited

Posts posted by livb89

  1. I made an object slowly rotate using llTargetOmega in the state_entry block 

    Then with it still rotating I did 'take' on the object to put it in my inventory.

    But now if I rez a copy without any script in it at all it still rotates.

    So it seems this rotation is now built into the object without any script. Is that normal?

  2. On 9/4/2019 at 9:50 PM, Alyona Su said:

    LL Viewer (and Kokua, I think) has the HDPI thing fixed

    It has been working fine for some weeks in the LL viewer. There is a bug in the current version of Firestorm (6.2.4) but it should be fixed in the next release because the issue has been resolved in Jira – see here https://jira.firestormviewer.org/browse/FIRE-24013 . I tested a 6.2.5 nightly release a few weeks ago and it was working well.

  3. On 9/4/2019 at 7:24 PM, livb89 said:

    I will see how it goes and write back here if it seems to be a fix or not.

    Just to update on this... I have found the problem is a lot better with the Viewer Texture Memory Buffer reduced to about 2/3 of its maximum. In my case it is now set to about 680 out of 1024. With is set like this I am finding that those arbitrary drops in frame rate in situations like the one I shared in my original post are no longer happening.

    The frame rate still drops of course with too much complexity but it is not doing it in situations where I would not expect it to (like above). So this is a good outcome!

  4. 3 hours ago, Alyona Su said:

    If you are on a 2015-2017 Mac - your best bet is windowed mode

    Interestingly I am finding Firestorm gives me a few more FPS when in full screen compared to being windowed (unless the window is really small). Mayne the system gives more resources to full screen apps.

    I have not tested this with HiDPI support switched on because there is a bug with it. But in the next release that seems to be fixed (I was one of the people who reported the bug, and they sent me a nightly on Jira with a fix)

  5. Someome on Twitter has suggested that in my case I try lowering that Viewer Texture Memory Buffer setting. They said:

    Quote

     

    That slider tells the client how much GPU memory it can use. If your system is using a lot of GPU memory already, SL may not have much to work with.

    Trying to use too much memory will mean having to swap data between system memory and GPU memory, which slows things down.

     

    On my iMac this Viewer Texture Memory Buffer setting only goes up to 1024 which is weird because on my MacBook Pro it goes to 2048. But the MacBook pro has a much worse integrated graphics device and also has half the RAM. The laptop is much slower overall for SL but it does not have the slowdown problem. It is just consistently struggling with the fans whirring away like jet engines a lot of the time!

    Anyway, I did a test earlier with this setting reduced from its 1024 maximum to about 640, and over about 30 minutes I did not notice any slowdowns. But I did notice problems like a picture on my wall had to load again after I cammed away from it and back again. Still.. this is better than the frame rate dropping down.

    I will see how it goes and write back here if it seems to be a fix or not.

  6. Hello :) I am running Firestorm 6.2.4  on an iMac. This is my computer information from the 'about' tab in Firestorm...

    CPU: Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz (3200 MHz)
    Memory: 32768 MB
    OS Version: Mac OS X 10.14.5 Darwin 18.6.0 Darwin Kernel Version 18.6.0: Thu Apr 25 23:16:27 PDT 2019; root:xnu-4903.261.4~2/RELEASE_X86_64 x86_64
    Graphics Card Vendor: ATI Technologies Inc.
    Graphics Card: AMD Radeon R9 M390 OpenGL Engine

    This graphics card has 2GB memory. I know that is not not the best for 'gaming' stuff, but it's a fast computer.

    I have attached screen shots of the graphics settings I use most of the time when I am alone at home or not in a busy place.

     

    In summary I am finding that the frame rate will start from a lovely 60-70 fps as a scene loads then settle at about 40 fps once loaded. Then, with the camera fixed on that view, a few minutes later the rate will usually fall down to about 5 fps.

    This is despite no avatars having entered the scene, and when I am (say) looking at a an area of my living room about the width of a sofa.

    Then a minute or so later it will climb up to about 40fps again. Or... if I cam around to a different angle that usually triggers it to speed up again too.

    ....

    Something that seems weird is that it is running faster when it has having to load lots of stuff in. For example I went on walk along a path next to all the houseboats in Belliseria and the whole time it is loading in the houseboats as i approach them, and during this walk the frame rate would be very high, say about 40fps. But then if i stop for a few minutes in one place, after everything has loaded the frame rate will quickly fall to about 4-5 fps

    This seems back to front – on my old laptop it was always slower when things were loading in, and faster once stuff was cached.

    When it has these slow down events the lag meter shows red for Client and says Possible cause: Images loading

    I have tried looking at the statistics bar but I can't really make much sense of it. However I have attached one screenshot of when it is fine then one from a minute taken in the same place after it dropped down.

    The other thing to mention is that this never happens with Advanced Lighting Model turned off.

     

    But just to reiterate - I can even walk around with the settings on Ultra and get about 25-30 fps. It is only when I stop in one place that it gets slow again.

    ...

    - I am wondering if there is something I could try changing in the settings? It seems like maybe a problem with the cache getting too full but I have cleared it and also tried setting it to the maximum size of 9984MB.

    - I have also done a complete reinstall which involved wiping the cache folders out but the problem is still here.

    Is there anything I can try? 

    Screenshot-2019-09-03-15.50.15.png

    Screenshot-2019-09-03-15.53.46.png

    Screenshot 2019-09-03 23.02.31.png

    Screenshot 2019-09-03 23.03.58.png

  7. 5 hours ago, Lucia Nightfire said:

    Since it was not mentioned, you can also use llSetLinkMedia() to store up to 27KB of data in 1KB blocks per legacy prim using the current, home and whitelist params.

    Caveats:

    • URL schema enforcement exists for current and home params. ("https://" added at the beginning of data if "http://" or "https://" is not present)
    • When using the whitelist param, there is a 64 count limit if data is in CSV format. (attempting to exceed it yields a script error)
    • Reducing face count results in data loss on faces that no longer exist.

    So if I understand this correctly this is storage is *designed* store media URLs but here we are doing a sort of a hack where we are using those fields to store other data as a string? Basically taking advantage of the fact that LL allow up to 1KB for each field (about 1024 characters? allowing for 'https://')

    Is that right?

  8. @Fritigern Gothly

    I'd not heard of strided lists, that's good to know. and I have sen there is llList2ListStrided() for getting data out too.

    Or yes... some known string separator that won't appear in the data itself. Lots of ways of doing this, thanks for the suggestions :)

     

     

    • Like 1
  9. 3 hours ago, Nova Convair said:

    and I don't need a timestamp

    The timestamp in my concept is more for literally recording the time of entry of the user in order to inform the owner. Not for list management purposes.

    • Like 1
  10. 4 hours ago, Rolig Loon said:

    Try storing part of the UUID and an abbreviated timestamp as a single list element  ( time + "~" + UUID, maybe). Then periodically -- every few hours, every week, whatever ... -- delete list elements with the oldest timestamps.  Those will always be the ones at the start of the list.

    It's a great idea, thanks. I will figure out some sort of string-based storage format like this, given that we can't make lists of complex objects like we can with arrays in javascript (I am a web designer in RL so know my way around that pretty well). 

  11. 3 minutes ago, Wulfie Reanimator said:

    What kinds of operations does it need do on that list?

    I am thinking it will record which avatars are present (checking say every 60 seconds) then it will record when they were there and possibly how long for. So a list of keys and maybe a parallel list of timestamps.

    The owner will be able to click to see the list. Basically a security system of sorts with a greeter built in. A more polite system than a security orb, kinda. Anyone on our whitelist gets a more welcoming message :)

  12. Thanks everyone - super helpful and so quickly!

    I won't go through and quote reply everyone but I have taken all that in. I think keeping them in the script memory is fine for our purposes.

    I kind of forgot that even when we log out the scripts keep running. I assume region restarts would cause them to be reset though?

    Maybe I will look into storing KVPs with an experience as an upgraded version once I get the basic thing running. And I will certainly look into greeters! I know this thing probably already exists but I want to write my own so I can learn better.

    • Like 1
  13. I am thinking of writing a script that records who visits our parcel then we know who has been there. Possibly then the device could speak to them (in nearby) if they are not on a whitelist. But also we don't want it to repeatedly say the same thing as that would be annoying.

    I got as far as getting the list and counting it using this:

    list avatarsInRegion = llGetAgentList(AGENT_LIST_PARCEL, []);
    integer numOfAvatars = llGetListLength(avatarsInRegion);

    Now I have done that I hoped to go through the avatars and record them. But I have just discovered it is not possible to write to a notecard, only read.

    Is there any SL way of recording information like this?

    The only other way I can think is with an external web service / API but then that all gets a lot more complicated to set up.

  14. 4 minutes ago, Fionalein said:

    answered above...

    Yes thanks, I saw that too. I was just responding to the specific point about 'read the notecard' ... which itself (the notecard) assumes this knowledge on the part of the person using the kit

    I assume they (makers of Maitreya) are coming at it with the assumption of people knowing how to make textures from the world before mesh bodies came along. But some of us are quite new to SL, never used a system avatar (other than for a few days as a noob), and never made any clothing the "old way".

    It is not actually that obvious (with no prior knowledge) that these avatar UV textures would work at all with mesh bodies. I had assumed for example that a different template would be needed for each of the tattoo, underwear and clothing layers because they are slightly different sizes. But I am going to give the template in that Robin Wood link a try now :)

  15. On 3/20/2019 at 10:27 PM, Mitzie Mohr said:

    don't understand something basic.  Then instructions for the Applier are to "drag the script into the prim."  What Prim?  I'm not clear on what that means.  To me a prim is one of the basic building blocks in SL.  You create it on your land.  What prim is this referring to?

    You need to make that prim. So yes - as in one of the basic building blocks - made with the build tool. Typically you would probably resize this into a flat rectangle and then attach it to your HUD. Then maybe add graphics to it so it looks nice. But basically when you click it the script will be triggered and apply the textures originally defined in the notecard.
     

  16. On 5/30/2018 at 12:47 AM, Coby Foden said:

    Did you read that notecard?

    I copy part of the instructions from the notecard here:
    1. Make sure to set the script "Maitreya Applier Script - V2 0.10" to NO TRANSFER, If you plan on selling your appliers!
    2. Drag the script "Maitreya Applier Script - V2 0.10" in to your prim.
    3. Fill in your texture UUID in the notecard "Maitreya Texture Loader - Body Overlay V2", then drag the notecard in to the prim as well.

    The point seeming to be missed as I understand it is that the dev kit instructions don't give guidance on how to make (or where to get) a template for the actual texture itself. The technical instructions themselves are absolutely fine but it is assumed that users know how to make the texture asset.

    It would be useful if they provided a template (or templates) with regions of the body marked so that people can then take this into Photoshop, remove the guides and add their own pixels.

  17. 13 hours ago, Alyona Su said:

    MY settings are (basically)

    Thanks, yes I have four presets for different purposes such as: parties, shopping, exploring, photography and so on, each with different draw distances, some with advanced lighting off and so on.

    I will give the LL viewer another go. I don't need RLV :)

  18. 3 hours ago, Alyona Su said:

    This is an LL Viewer feature (and it looks amazingly gorgeous when turned on - Windows can't even come close, much less touch it) - I imagine Firestorm, Catznip (if they ever finally release a macOS version,) Kokua and the rest will implement the technology when they are able or, at least, they'd be smart to do so.

    Oh OK. The reason I asked is because it looks as though Firestorm on my retina Macbook Pro is already using the full display resolution. I mean... if I take a screenshot with Firestorm full screen it is 3360 × 2100 and the text looks pretty sharp. Maybe I am wrong.

    I was actually looking see if I could turning the Hi DPI setting *off* in order to see if I could get a better frame rate. While SL looks great on my retina display I struggle to get about 10 FPS in some places, where my partner who has a 1920x1080 72 dpi display on a PC gets more like 20 FPS at the same place. There are other factors... mine is a laptop so not the best graphics card. But I was just experimenting with performance generally.

    (And yes i know about changing other graphics settings like draw distance, max complexity, and shadows, reflections and so on - I already have presets)

  19. On 4/26/2018 at 5:58 PM, Chic Aeon said:

    The problem is like many to most Blender tutorials, this one (I watched it long ago) was about designing a ring to be used JUST as a RENDER and not baked (so you get a pretty photo of a ring and you have that).

    Oh ty so much! I was wondering for so long when when I imported my collada file SL was saying there was no texture to import with it. I have never even heard of "baked". I think I have been doing the wrong Blender tutorials. So much more to learn always! :)

    • Like 1
  20. I tried this with Blender 2.79 and after you pick the 'SL+Open Sim Static' Operator Presets you have to then choose between either UV textures or Materials because that is now a dropdown and not two separate checkboxes. But I have tried both options.

    I can see my mapped texture in Blender. But when I go to import the DAE file in SL it says there are no textures after I click calculate costs.

     

    Screenshot 2019-06-02 21.50.55.png

  21. 1 minute ago, Abnor Mole said:

    Even if you put it 100m in the air with a range of 400m, it still only works to 400m above sea level. The warnings are in a chat message and not a popup. As long as you don't stop over a parcel for more than a few seconds, you shouldn't notice anything but the occasional chat message warning (which can be easily ignored).

    Oh that sounds fine then. Hopefully LL won’t allow pop ups from third party orbs either.

×
×
  • Create New...