Horrible seams suddenly

Anyone else having trouble with simple prim alignments lately? I saw another thread about this around 05/14, but it did not get resolved.

I'm having trouble with what appears to be noticeable gaps in all my seams suddenly as I am building. I build by numbers or I use prim docker. Neither improves the situation.

Older builds I have made still look fine. Is this some kind of prim drift or an instability issue? Maybe I should stop trying to build until I find out what's going on.

I've been running into this too. I'm building using bright, solid-colored blocks, so the seams are horrifically visible and ugly. It's even worse on the semi-transparent parts, where there appear to be overlaps or gaps even though I know I placed them just right.

I've built my entire structure by the numbers, but none of it's lining up. Some stuff I built yesterday seems just fine, but it's all gone to hell today. I even try re-editing the objects (when I do, I notice the positions are off by 0.001-0.005) but the seams won't go away.

Here's an idea: a "snap" tool in SL. Something where you can click on a prim or group of linked prims, and its X,Y,Z coords snap to the nearest 0.01 or something.

http://sl.elitehost.org/images/seams.gif
Look at my seams, and see how annoying they are (on a surface that's supposed to be solid yellow). They're even worse in-game, because they "sparkle" as you move, making them 10x more noticeable

I had the same problem .I tryed prim docker and manually by numbers still had seams that wouldnt go away.Then I noticed the rotation on some of my prims was out very slightly even thou I hadnt set rotation Im not sure if it was havok 4 that had done it but that was why they wouldnt align.

Yes, Phil. Your picture is exactly what I'm seeing. I wonder if it 's Windlight causing this, but why on some builds and not on others? I may try looking in different environment settings.

I'll also try looking at the rotation. I didn't think of that.

Thanks.

Hmmm, my prim specs are as follows, and I still get seams:

10.000x10.000x0.500 @ <65.000,199.000,36.250>
10.000x10.000x0.500 @ <75.000,199.000,36.250>
10.000x10.000x0.500 @ <85.000,199.000,36.250>
10.000x10.000x0.500 @ <65.000,209.000,36.250>
10.000x10.000x0.500 @ <75.000,209.000,36.250>
10.000x10.000x0.500 @ <85.000,209.000,36.250>

all have 0.000 rotation on every axis.

/me pokes SL with a stick and grumbles to himself

SL has always had problems with seams. I really don't like them either

I've worked really hard on my builds to reduce their visibility.

i also have noticed effects like sudden little rotations or minimal offsets appearing out of nowhere. something just jumps up in my mind, which may be the cause of this effect:
i recently was sitting in front of a computer with a space navigator connected. While working i noticed, that suddenly my object was jumping away sideways. I realised, that the space navigator was active and reacted to my moves on the table, although i didn't touch it. So i realised, that you can move your objects around with that device. This seems to be the default configuration. Maybe the effect you describe disappears, when you disable 3D mouse in build mode ? Even if you do not have a 3D mouse connected ?

I have also noticed a sudden jump in the number of naturally occurring seams. Snap to grid and pure numbers seem to generate them very frequently now - which can be infuriating.

There are three possible points of failure you should be looking out for: coordinates, scale and rotation. Any of these properties could be off by a tiny amount and result in a seam. Rotation especially if you use the shift+drag copy method.

I've been using a script that reports these values to a higher precision to see what is off. While tedious, it had helped to squash a lot of the seams I encounter. Never used Prim docker, so I'm not sure if that's the same thing.

What script are you using Fenix?

My guess is it's mostly a client issue. I normally use Nicholaz, which is based on a several-months-old viewer, and I do not have the seams issue. If I run the latest official viewer I see seams (sometimes even gaps) in existing builds where they never were before. It can make a professional build look like amateurish crap, and it's not even the builder's fault.

Have you tried using Nicholaz or some other viewer to determine if the seams are still there? If they aren't, that would confirm it's a client issue, and it would be easier to Jira.

Hmm, I never tried the other viewers - but perhaps now I should look into Nicholaz's

As for my script, it's just something I literally hacked together (mostly copied) from the LSL wiki examples. It reports by simply saying llGetScale(), llGetPos() and the llGetRot() (though the rotation requires some conversions). Setting the values is also pretty straight forward, and I created my own functions to speed it things up a tad. I can send you the script if you want.

vectors are three 32bit floats. even if you use the grid the resulting position float can be off by 0.0005 without appearing in the editor? because it cuts/rounds the REGION position, range [0 ... 256[, after .000 digits!
southwest parcels are more accurate than northeast parcels?
usually the inaccuracy is too small to be noticeable.

also very inaccurate z-buffering.

The thing about building is - I think you should use the default viewer and default atmosphere settings because that's what most people who see your build will be using. I'm to the point where I'll just put it down to SL being buggy, because other builds look just as bad.

It's highly frustrating though and makes me want ot just go socialize instead.

I came to the forum to report something similar happening here, and there's already a thread. Ah, good, it's not just me then!

I can't see how the client can make any difference at all in this case (using RC9 in all cases) - I have a parasol I'm building which has a canopy consisting of a central ring with eight truncated triangular prims around it, arranged radially in a daisywheel manner. The inner edges of the outer prims exactly touch the ring in exactly the right place so they line up neatly. All nine prims are linked, with the ring being the root prim.

I did it all by numbers, both positions and rotations, with the three-decimal accuracy the editor allows, and it looked perfect. I shifted the link set to incorporate it into the rest of the build, and as I was lining it up I noticed by eye that the set was no long longer properly lined up. On checking the numbers, the outer prims were fine relative to one another, but the central ring had shifted. WTF?! There's no way I could have accidentally moved just one of the prims of the set, surely? I'd remember if I'd selected "Edit linked parts", surely? Maybe not, maybe it was momentary insanity.

Sooo, unlinked the set and and repositioned all the prims, checking very carefully that everything lined up precisely before relinking and rechecking. This was last night. Today, when I came to resume work, the central prim had again shifted, by 0.02m in the X axis.

Here's a snapshot to give you an idea: http://tinyurl.com/3jl9ma

You can see that the central ring is at X=228.480; it should be at X=228.500 like the top and bottom outer prims in the pic (and it was last night). You can see the gap this has left on the right, and how's it's closed up on the left - this is a show-stopper for this build, it looks so badly-done. Bah!

And all those times in the past I've blamed myself for poor lining-up, maybe at least some of the time it wasn't me. And if it's affecting normal-scale building like this, how about those poor people trying to make jewellery?!

There are a few unrelated issues that can have similar symptoms. The new client does create seams and gaps where none were before (at least on my system and some others I've heard about), but there are also occasional (long-standing) problems with prim alignment being incorrect because the UI doesn't show or let you edit to correct precision, and sometimes prims move a bit for unknown reasons.

There is a Jira on this I'll try and find it, plus also the copy shift function with snapping to grid then linking will *move* your alignments - the act of linking makes the prims slightly unhappy. That may have gone into the latest RC by now, not sure.

Will report back if I can find the details.

when using a sphere or circle next to another prim so that there is a seam, try replacing it with a torus, if it still works for the design of course. i found I eliminated some of the 'gappy seam' effect one gets from looking at a distance.