# Video Enhance AI v1.8.1

**URL:** <https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807>\
**Category:** Releases\
**Created:** [December 22, 2020, 9:26pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807 "2020-12-22T21:26:46Z")\
**Posts on this page:** 20\
**Page:** 11

<div class="post-metadata">

**Author:** ![pov](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/p/eb8c5e/32.png) [@pov](https://community.topazlabs.com/u/pov)\
**Post date:** [January 3, 2021, 9:47pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/205 "2021-01-03T21:47:51Z")

</div>

> [@deepzone](#):
>
> Been using Gaia v6 a bit now. And i think it’s better than v5 in some cases.

Where/how did you get Gaia v6? It’s not in this release unless I am missing something.

---

<div class="post-metadata">

**Author:** ![mxrevolution](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/m/c89c15/32.png) [@mxrevolution](https://community.topazlabs.com/u/mxrevolution)\
**Post date:** [January 3, 2021, 10:59pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/206 "2021-01-03T22:59:34Z")

</div>

i am also running v1.8.1 … and there is no gaia v6

---

<div class="post-metadata">

**Author:** ![lee.keels](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/l/a88e4f/32.png) [@lee.keels](https://community.topazlabs.com/u/lee.keels)\
**Post date:** [January 4, 2021, 1:29am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/207 "2021-01-04T01:29:55Z")

</div>

If you kept those models, you can use them in 1.8.1.

---

<div class="post-metadata">

**Author:** ![Chase](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/c/76d3ee/32.png) [@Chase](https://community.topazlabs.com/u/Chase)\
**Post date:** [January 4, 2021, 3:18am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/208 "2021-01-04T03:18:35Z")

</div>

> [@pov](#):
>
> Where/how did you get Gaia v6? It’s not in this release unless I am missing something.

_Edit - corrected path_

You have to add it manually.

First, close VEAI.

Open this file:

`C:\Users\[username]\AppData\Roaming\Topaz Labs LLC\Video Enhance AI\models\gaia-hq-5.json`

Change line 3:

` "version": "5",`

to

` "version": "6",`

Now, save the file as:

`C:\Users\[username]\AppData\Roaming\Topaz Labs LLC\Video Enhance AI\models\gaia-hq-6.json`

Now when you start VEAI, both Gaia v5 and Gaia v6 should appear in your model list. The program will download the files it needs as needed.

---

<div class="post-metadata">

**Author:** ![gaugaugaming2](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/g/eb9ed0/32.png) [@gaugaugaming2](https://community.topazlabs.com/u/gaugaugaming2)\
**Post date:** [January 4, 2021, 3:48am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/210 "2021-01-04T03:48:28Z")

</div>

I have no problem with NVENC qp0 in ffmpeg (consider it’s lossless, 99.9%), output is even better than CPU encoding (using my poor eyes 🤣). I don’t encode at target rate, only in rare occasions due to limitations. Same result in both H264 and H265.

---

<div class="post-metadata">

**Author:** ![thevance2000](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/t/6a8cbe/32.png) [@thevance2000](https://community.topazlabs.com/u/thevance2000)\
**Post date:** [January 4, 2021, 10:31am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/211 "2021-01-04T10:31:08Z")

</div>

What’s better 3090 or 6900xt?

---

<div class="post-metadata">

**Author:** ![jphilip-649517](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/j/a587f6/32.png) [@jphilip-649517](https://community.topazlabs.com/u/jphilip-649517)\
**Post date:** [January 4, 2021, 2:23pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/212 "2021-01-04T14:23:27Z")

</div>

You want lossless only in the case of further editing in a NLE. And what’s worse than GOP based codec for editing ? Or I hope your qp=0 NVENC files are at least --keyint=1 also.  
Better work with real intraframe lossless codecs like MagicYUV.  
And the dev team should better consider x265 (that also can be lossless) than NVENC. Or sudenly will VEAI will magically be enable to compute 10x faster just because it is encoding with NVENC ?  
No, you will get the same speed and in that case a normal brain would choose the better output quality solutions with other crf values.  
NVENC is just a terrible codec. I’ve seen 4K H265 Nvenc encoded files even @35MB/s with awful banding artifacts. NVENC is meant for speed and live streaming purpose.

---

<div class="post-metadata">

**Author:** ![raymond.monaghan](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/r/a87d85/32.png) [@raymond.monaghan](https://community.topazlabs.com/u/raymond.monaghan)\
**Post date:** [January 4, 2021, 3:48pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/213 "2021-01-04T15:48:16Z")

</div>

> [@thevance2000](#):
>
> What’s better 3090 or 6900xt?

At the moment it’ll be neither, buying one of the new GFX cards for Topaz would be a big waste of your cash. It’ll take a bit of time before Topaz and other software is fine-tuned to take full advantage of these new architectures.

---

<div class="post-metadata">

**Author:** ![Red](https://community.topazlabs.com/user_avatar/community.topazlabs.com/red/32/82556_2.png) [@Red](https://community.topazlabs.com/u/Red)\
**Post date:** [January 4, 2021, 6:22pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/214 "2021-01-04T18:22:48Z")

</div>

Whichever is in stock I suppose.

---

<div class="post-metadata">

**Author:** ![pkatmd-junk12](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/p/2acd7d/32.png) [@pkatmd-junk12](https://community.topazlabs.com/u/pkatmd-junk12)\
**Post date:** [January 5, 2021, 6:10am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/215 "2021-01-05T06:10:09Z")

</div>

Hmm, my “gaia-hg-5.json” file is located here instead:

C:\Users\\<your\_username\>\AppData\Roaming\Topaz Labs LLC\Video Enhance AI\models

And I have not noticed any differences in the quality of v5 vs v6…

---

<div class="post-metadata">

**Author:** ![pkatmd-junk12](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/p/2acd7d/32.png) [@pkatmd-junk12](https://community.topazlabs.com/u/pkatmd-junk12)\
**Post date:** [January 5, 2021, 6:34am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/216 "2021-01-05T06:34:31Z")

</div>

Interestingly, I haven’t noticed any speed differences during my tests with either my RTX 2070 (non-Super) vs my RTX 3070. I haven’t tried my GTX 1060 (3GB) yet due to lack of time.

My take: At this point, don’t waste your money. I think it would make more sense to invest in a faster CPU (like a 3950X) especially if you are using Staxrip to perform pre-processing cleanups before you feed your videos to VEAI. If scripted properly, Staxrip is able to push my 3950X to nearly 100% CPU usage.

---

<div class="post-metadata">

**Author:** ![tabnk](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/t/9d8465/32.png) [@tabnk](https://community.topazlabs.com/u/tabnk)\
**Post date:** [January 5, 2021, 7:45am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/217 "2021-01-05T07:45:30Z")

</div>

Tried Gaia V6. Notice some improvement on the facial department. Less weirdly look than before but still got room for improvement.

---

<div class="post-metadata">

**Author:** ![lee.keels](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/l/a88e4f/32.png) [@lee.keels](https://community.topazlabs.com/u/lee.keels)\
**Post date:** [January 5, 2021, 10:45am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/218 "2021-01-05T10:45:34Z")

</div>

> [@pkatmd-junk12](#):
>
> C:\Users\<your\_username\>\AppData\Roaming\Topaz Labs LLC\Video Enhance AI\models

There’s literally no difference in yours and Chase’s.

---

<div class="post-metadata">

**Author:** ![mxrevolution](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/m/c89c15/32.png) [@mxrevolution](https://community.topazlabs.com/u/mxrevolution)\
**Post date:** [January 5, 2021, 7:38pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/220 "2021-01-05T19:38:22Z")

</div>

i assumed that …

but i have never seen gaia V6 … from 1.6.1 to 1.8.1 there was no v6

---

<div class="post-metadata">

**Author:** ![deeImgGuy](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/3da27b/32.png) [@deeImgGuy](https://community.topazlabs.com/u/deeImgGuy)\
**Post date:** [January 5, 2021, 7:45pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/221 "2021-01-05T19:45:49Z")

</div>

It was in the 1.8.0 beta. Higher versions of Theia were also there; the assumption is that they found problems greater than with earlier versions and so rolled back.

---

<div class="post-metadata">

**Author:** ![Hussey](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/h/dec6dc/32.png) [@Hussey](https://community.topazlabs.com/u/Hussey)\
**Post date:** [January 6, 2021, 12:10am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/222 "2021-01-06T00:10:35Z")

</div>

Actually, photoshop and a number of other graphics software DO let you keep older versions of how their software functions, and let the users decide if and when they want to use the old methods.

---

<div class="post-metadata">

**Author:** ![lee.keels](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/l/a88e4f/32.png) [@lee.keels](https://community.topazlabs.com/u/lee.keels)\
**Post date:** [January 6, 2021, 2:51am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/223 "2021-01-06T02:51:54Z")

</div>

Please show an example.

---

<div class="post-metadata">

**Author:** ![pkatmd-junk12](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/p/2acd7d/32.png) [@pkatmd-junk12](https://community.topazlabs.com/u/pkatmd-junk12)\
**Post date:** [January 6, 2021, 7:14am UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/224 "2021-01-06T07:14:05Z")

</div>

???

Yes, there is _literally_ a difference!

If you don’t include the in your directory (especially when your computer is being used by several users), you won’t find the Topaz Labs LLC models!

BTW, Chase edited his post to correct his entry _after_ my post – in case you were wondering.

---

<div class="post-metadata">

**Author:** ![vico4444](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/v/bc79bd/32.png) [@vico4444](https://community.topazlabs.com/u/vico4444)\
**Post date:** [January 6, 2021, 12:38pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/225 "2021-01-06T12:38:23Z")

</div>

Hello, newbie here, this software is amazing but it would be better if you could add a “select region” option

---

<div class="post-metadata">

**Author:** ![TPX](https://community.topazlabs.com/user_avatar/community.topazlabs.com/tpx/32/194727_2.png) [@TPX](https://community.topazlabs.com/u/TPX)\
**Post date:** [January 6, 2021, 5:02pm UTC](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807/226 "2021-01-06T17:02:08Z")

</div>

There are functions in PS where you can mark “use old version”.

But that not that much, thats only a legacy Brush and a Contrast function.

But, let’s say if many users want to have a certain model again and it works with the new engine then you could make it downloadable.

[Previous page](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807.md?page=10)

[Next page](https://community.topazlabs.com/t/video-enhance-ai-v1-8-1/18807.md?page=12)
