# The anti-alias quality between Gmask and GmaskTracer

**URL:** <https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462>\
**Category:** Flame Questions\
**Tags:** action, batch\
**Created:** [May 26, 2023, 7:37am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462 "2023-05-26T07:37:51Z")\
**Posts on this page:** 20\
**Page:** 1

<div class="post-metadata">

**Author:** ![JeffChin](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/jeffchin/32/3670_2.png) [@JeffChin](https://forum.logik.tv/u/JeffChin)\
**Post date:** [May 26, 2023, 7:37am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/1 "2023-05-26T07:37:51Z")

</div>

Hi, everyone:

When comparing the result of the same mask shape with the same edge softness offset (set to 1.0) from these two different masking tools, the old GMask generate less aliasing result but the almighty GMaskTracer shows those zig-zag aliasing edge when an clean and sharp edge of masking is expected…

Have tried several ways such as larger softness value and tighten up by MatteEdge; AXFF; forced 2DTransform Filter but only worsen the result.

Is it the consequence I need to face after betray my ex lover Gmask? 😜

 ![GMask_vs_GmaskTracer](https://us1.discourse-cdn.com/flex020/uploads/thedepartmentofexternalservices/original/2X/8/8ad0d3906983e3c2c15a8ac28bf5fddad078a898.jpeg)

---

<div class="post-metadata">

**Author:** ![PlaceYourBetts](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/placeyourbetts/32/6852_2.png) [@PlaceYourBetts](https://forum.logik.tv/u/PlaceYourBetts)\
**Post date:** [May 26, 2023, 9:03am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/2 "2023-05-26T09:03:29Z")

</div>

Have you been into the node settings of gmaskTracer and upped your sampling?

Total game changer.

---

<div class="post-metadata">

**Author:** ![JeffChin](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/jeffchin/32/3670_2.png) [@JeffChin](https://forum.logik.tv/u/JeffChin)\
**Post date:** [May 26, 2023, 9:49am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/3 "2023-05-26T09:49:44Z")

</div>

Aha~there it is! Thanks lot !! You’re my weekend changer~  
I thought my default AA 4 was enough but actually need to upped to 12+ for this case. Now I can totally embrace my new lover 😅

---

<div class="post-metadata">

**Author:** ![BrittCiampa](https://avatars.discourse-cdn.com/v4/letter/b/ec9cab/32.png) [@BrittCiampa](https://forum.logik.tv/u/BrittCiampa)\
**Post date:** [May 26, 2023, 3:05pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/4 "2023-05-26T15:05:14Z")

</div>

Doesn’t this just massively slow you down though??? I try to never ever ever use AA filtering on anything in flame if I can avoid it. However, if I’m missing something, I’m down to change my mind!!

---

<div class="post-metadata">

**Author:** ![hBomb42](https://avatars.discourse-cdn.com/v4/letter/h/bbce88/32.png) [@hBomb42](https://forum.logik.tv/u/hBomb42)\
**Post date:** [May 26, 2023, 3:56pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/5 "2023-05-26T15:56:25Z")

</div>

I just do a shift+a-drag on the spline (not a vert) and add the teeniest, tiniest amount of feather to my mask and can run happily at 1 AA sample.

---

<div class="post-metadata">

**Author:** ![Jeff](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/jeff/32/9305_2.png) [@Jeff](https://forum.logik.tv/u/Jeff)\
**Post date:** [May 26, 2023, 4:04pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/6 "2023-05-26T16:04:00Z")

</div>

I’m with @hBomb42 on the slight softness before upping the sample rate. I don’t exit the gmask tracer without adding just a touch of softness for everything I do.

---

<div class="post-metadata">

**Author:** ![andy\_dill](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/andy_dill/32/47_2.png) [@andy\_dill](https://forum.logik.tv/u/andy_dill)\
**Post date:** [May 26, 2023, 5:32pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/7 "2023-05-26T17:32:32Z")

</div>

@dlevine pointed this out to me, and in the tests the old gmask had a nicer edge even with the antialiasing way up.

As a gmask tracer evangelist I felt defeated.

---

<div class="post-metadata">

**Author:** ![BrittCiampa](https://avatars.discourse-cdn.com/v4/letter/b/ec9cab/32.png) [@BrittCiampa](https://forum.logik.tv/u/BrittCiampa)\
**Post date:** [May 26, 2023, 5:52pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/8 "2023-05-26T17:52:53Z")

</div>

100% this. If you’re using AA samples as opposed to just the tiniest feather you are substantially increasing render time. Maybe doesn’t feel like a lot on a simple mask, but it becomes a cumulative problem. I’ve seen batches using AA all over the place and I can get the same result with some feathering, some 1pixel blurs, and surface filtering (like EWA+Linear) and render that thing in half the time. At least.

---

<div class="post-metadata">

**Author:** ![johnt](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/johnt/32/14040_2.png) [@johnt](https://forum.logik.tv/u/johnt)\
**Post date:** [May 26, 2023, 6:27pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/9 "2023-05-26T18:27:51Z")

</div>

Yep. That and the splining beauty of the old gmask means it’s still relabvant.

---

<div class="post-metadata">

**Author:** ![PlaceYourBetts](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/placeyourbetts/32/6852_2.png) [@PlaceYourBetts](https://forum.logik.tv/u/PlaceYourBetts)\
**Post date:** [May 26, 2023, 7:41pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/10 "2023-05-26T19:41:21Z")

</div>

I hear all your concerns and they are totally valid. This is what I do for a hard edge roto.

Draw my shape. Tiny gradient out from my spline to match the soft pixels on the edge and then samples to +4 to make it look nice.

Never noticed any significant slow down.

---

<div class="post-metadata">

**Author:** ![BrittCiampa](https://avatars.discourse-cdn.com/v4/letter/b/ec9cab/32.png) [@BrittCiampa](https://forum.logik.tv/u/BrittCiampa)\
**Post date:** [May 26, 2023, 7:50pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/11 "2023-05-26T19:50:01Z")

</div>

Sure, at AA 4 it’s not gonna be very noticeable. But things pile up. And I am probably too paranoid about making renders efficient. It’s almost a game at this point. And… I like to go faaaaast LOL. But above, talking about AA sample values of 12+, that’s going to cause a significant slowdown… combine that with motion blur, you’re going to lunch and hoping the track doesn’t slip.

---

<div class="post-metadata">

**Author:** ![PlaceYourBetts](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/placeyourbetts/32/6852_2.png) [@PlaceYourBetts](https://forum.logik.tv/u/PlaceYourBetts)\
**Post date:** [May 26, 2023, 7:53pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/12 "2023-05-26T19:53:53Z")

</div>

I have a similar thing when I use the Text tool.

I never except the Auto Softness. I usually make type 16 samples. But then you can usually freeze a mux after that 😉

---

<div class="post-metadata">

**Author:** ![PlaceYourBetts](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/placeyourbetts/32/6852_2.png) [@PlaceYourBetts](https://forum.logik.tv/u/PlaceYourBetts)\
**Post date:** [May 26, 2023, 7:59pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/13 "2023-05-26T19:59:42Z")

</div>

> [@BrittCiampa](#):
>
> hoping the track doesn’t slip.

Ha. My track won’t Slip because I have access to the planar tracker (and motion vectors) in the gmaskTracer 😉

---

<div class="post-metadata">

**Author:** ![BrittCiampa](https://avatars.discourse-cdn.com/v4/letter/b/ec9cab/32.png) [@BrittCiampa](https://forum.logik.tv/u/BrittCiampa)\
**Post date:** [May 26, 2023, 8:00pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/14 "2023-05-26T20:00:57Z")

</div>

Ha! I’m 100% team gmask tracer!!! This aint about that. It’s about using AA period on anything. It’s super slow and is a game of diminishing returns. The day I use the OG gmask is the day I turn autokey off\*…

\*NEVER

---

<div class="post-metadata">

**Author:** ![GPM](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/gpm/32/4224_2.png) [@GPM](https://forum.logik.tv/u/GPM)\
**Post date:** [May 26, 2023, 8:58pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/15 "2023-05-26T20:58:58Z")

</div>

Have you tried bumping HWAA instead?  
And of course autokey off 🙂

---

<div class="post-metadata">

**Author:** ![BrittCiampa](https://avatars.discourse-cdn.com/v4/letter/b/ec9cab/32.png) [@BrittCiampa](https://forum.logik.tv/u/BrittCiampa)\
**Post date:** [May 26, 2023, 9:04pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/16 "2023-05-26T21:04:04Z")

</div>

Aside from the occasional need for working with text, I simply don’t use anti-aliasing. 1000 different ways to skin a cat.

---

<div class="post-metadata">

**Author:** ![AdamArcher](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/adamarcher/32/3529_2.png) [@AdamArcher](https://forum.logik.tv/u/AdamArcher)\
**Post date:** [May 27, 2023, 5:15am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/17 "2023-05-27T05:15:05Z")

</div>

I haven’t noticed much of a slowdown with HWAA but now I’m going to have to test it.

---

<div class="post-metadata">

**Author:** ![PlaceYourBetts](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/placeyourbetts/32/6852_2.png) [@PlaceYourBetts](https://forum.logik.tv/u/PlaceYourBetts)\
**Post date:** [May 27, 2023, 5:22am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/18 "2023-05-27T05:22:22Z")

</div>

Unfortunately I noticed no improvement to the gmaskTracer when I upped the HWAA. This was the first thing I tried.

---

<div class="post-metadata">

**Author:** ![JeffChin](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/jeffchin/32/3670_2.png) [@JeffChin](https://forum.logik.tv/u/JeffChin)\
**Post date:** [May 27, 2023, 6:28am UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/19 "2023-05-27T06:28:52Z")

</div>

Thanks for all the input and experience sharing~

In my case, I found the manually Shift+A softness usually more than 1.0 (I’m old with trembling hand…) and a bit too soft so I the adjust the Offset slider to make sure the masking is tight enough.

AA Sample 4 is part of my preference for GMask Tracer and I save it in User Bin for daily routine, It’s fine for most of the time but issue comes when the shot has some round and sharp objects. In this case I wasn’t realized that sometimes need to upped to such a high value in order to match the quality from legacy GMask but sacrifice the render time.

I like new GMask’s Edit Tracking Keyframes and superior motion blur render, but not for it’s “default” render quality and, the most important thing to me, the “bluntness” feel when dragging those vertices!  
Curious if anyone fell the same as me about this??

---

<div class="post-metadata">

**Author:** ![andy\_dill](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.logik.tv/andy_dill/32/47_2.png) [@andy\_dill](https://forum.logik.tv/u/andy_dill)\
**Post date:** [May 27, 2023, 11:47pm UTC](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462/20 "2023-05-27T23:47:10Z")

</div>

> [@johnt](#):
>
> Yep. That and the splining beauty of the old gmask means it’s still relabvant.

I disagree, actually.

Despite making a nicer edge at low softnesses, I cannot think of a single time in the last ten years where that was the obstacle to my comp coming together well.

I’m not saying you can’t like it, but I do not miss it at all, and I’ll gladly take this issue in trade for all the improvements the new one has, especially the ability to draw my own gradients. My old setups were positively LITTERED with matte edge nodes.

[Next page](https://forum.logik.tv/t/the-anti-alias-quality-between-gmask-and-gmasktracer/8462.md?page=2)
