# Topaz Video 1.0.3 - Patch

**URL:** <https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907>\
**Category:** Releases\
**Created:** [October 27, 2025, 9:03pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907 "2025-10-27T21:03:00Z")\
**Posts on this page:** 20\
**Page:** 5

<div class="post-metadata">

**Author:** ![ArtistryInLight](https://community.topazlabs.com/user_avatar/community.topazlabs.com/artistryinlight/32/194879_2.png) [@ArtistryInLight](https://community.topazlabs.com/u/ArtistryInLight)\
**Post date:** [October 31, 2025, 8:03pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/83 "2025-10-31T20:03:09Z")

</div>

Just got new versions of all apps. 1.0.3 Crashes 99% upon exit on win11, only 1 time did it not crash. 3060 12gb, latest nv drivers. Runs fine otherwise..

---

<div class="post-metadata">

**Author:** ![kyle.topazlabs](https://community.topazlabs.com/user_avatar/community.topazlabs.com/kyle.topazlabs/32/203506_2.png) [@kyle.topazlabs](https://community.topazlabs.com/u/kyle.topazlabs)\
**Post date:** [October 31, 2025, 8:42pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/84 "2025-10-31T20:42:47Z")

</div>

The devs are aware and working on a patch. The support team has been trying to get through all of the forum posts to respond to everyone as quickly as we can.

---

<div class="post-metadata">

**Author:** ![kyle.topazlabs](https://community.topazlabs.com/user_avatar/community.topazlabs.com/kyle.topazlabs/32/203506_2.png) [@kyle.topazlabs](https://community.topazlabs.com/u/kyle.topazlabs)\
**Post date:** [October 31, 2025, 8:48pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/85 "2025-10-31T20:48:16Z")

</div>

Please email the support team directly so we can assist. [Help@topazlabs.com](mailto:Help@topazlabs.com)

The new Video app is not compatible with Intel-Macs and the previous Video AI is only compatible up to version 7.0.2.

---

<div class="post-metadata">

**Author:** ![SkyDude](https://community.topazlabs.com/user_avatar/community.topazlabs.com/skydude/32/241138_2.png) [@SkyDude](https://community.topazlabs.com/u/SkyDude)\
**Post date:** [October 31, 2025, 10:11pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/86 "2025-10-31T22:11:00Z")

</div>

But is the quality of the results the same? That is the main question… I have heard otherwise.

😋

---

<div class="post-metadata">

**Author:** ![jo.vo](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/j/f19dbf/32.png) [@jo.vo](https://community.topazlabs.com/u/jo.vo)\
**Post date:** [November 1, 2025, 8:47am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/87 "2025-11-01T08:47:06Z")

</div>

It’s roughly the same - apart from ever so often (re-)occurring bugs like the actual Starlight model on AppleSilicon which is totally unusable quality-wise or in the 1.0.2 the bug on NVidia where it was nearly impossible to get higher than 2x models working (and there AppleSilicon was much superior quality-wise).

BTW, NVIDIA would take much more profit from reducing accuracy than AppleSilicon as those chips do get much higher performance in e.g. FP8 or FP16 performance compared to FP32. AppleSilicon does not. So, apart from bugs I see absolutely no reason why NVIDIA should have higher quality.

---

<div class="post-metadata">

**Author:** ![jeremy.e](https://community.topazlabs.com/user_avatar/community.topazlabs.com/jeremy.e/32/102573_2.png) [@jeremy.e](https://community.topazlabs.com/u/jeremy.e)\
**Post date:** [November 3, 2025, 6:03am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/88 "2025-11-03T06:03:17Z")

</div>

Just wanted to make aware, for some reason all Starlight Sharp “Cloud” renders are only using 1X upscale it seems.  
No matter the resolution, it disregards 2x-3x-4x selections, as well as removes the audio in the export.  
The normal Starlight and the other models seem to work fine though.

- Jeremy Eckert.

---

<div class="post-metadata">

**Author:** ![MovingPixels](https://community.topazlabs.com/user_avatar/community.topazlabs.com/movingpixels/32/78554_2.png) [@MovingPixels](https://community.topazlabs.com/u/MovingPixels)\
**Post date:** [November 3, 2025, 9:02am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/89 "2025-11-03T09:02:47Z")

</div>

Just to add, same here. Gone back to 1.0.2 for now.

---

<div class="post-metadata">

**Author:** ![dianac](https://community.topazlabs.com/user_avatar/community.topazlabs.com/dianac/32/258112_2.png) [@dianac](https://community.topazlabs.com/u/dianac)\
**Post date:** [November 3, 2025, 4:54pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/90 "2025-11-03T16:54:46Z")

</div>

In every case I have investigated here (on Windows), Starlight Sharp upscales to some intermediate resolution, then uses Nyx (the denoise model) to upscale to your target format. For example, when I do a 3X upscale of a 640x480 original to 1920x1440 the Starlight process only produces a 1440x1080 output without any audio, then uses Nyx to upscale to 1920x1440. Not sure about regular Starlight.

I wish they wouldn’t do that Nyx process, or at least not delete the Starlight output, since I prefer to do any additional upscaling in Proteus. As it is, I grab the 1080p file from the work directory while the Nyx process is running and then terminate the run.

---

<div class="post-metadata">

**Author:** ![m.sanders](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/m/df705f/32.png) [@m.sanders](https://community.topazlabs.com/u/m.sanders)\
**Post date:** [November 3, 2025, 9:21pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/91 "2025-11-03T21:21:12Z")

</div>

**If I pause an export and then resume it:**

- Input color override settings are no longer applied from that point in the video
- Gets stuck when it starts “Finalizing…”

Exporting to ProRes 422 Std

MacBook M4 Max 64GB, MacOS Tahoe 26.0.1

---

<div class="post-metadata">

**Author:** ![ArtistryInLight](https://community.topazlabs.com/user_avatar/community.topazlabs.com/artistryinlight/32/194879_2.png) [@ArtistryInLight](https://community.topazlabs.com/u/ArtistryInLight)\
**Post date:** [November 4, 2025, 11:41am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/92 "2025-11-04T11:41:25Z")

</div>

Seems to be not crashing upon exit now.

---

<div class="post-metadata">

**Author:** ![jam7000](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/j/d2c977/32.png) [@jam7000](https://community.topazlabs.com/u/jam7000)\
**Post date:** [November 4, 2025, 1:48pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/93 "2025-11-04T13:48:49Z")

</div>

can both starlight models be added to benchmark in next update please?

---

<div class="post-metadata">

**Author:** ![kyle.topazlabs](https://community.topazlabs.com/user_avatar/community.topazlabs.com/kyle.topazlabs/32/203506_2.png) [@kyle.topazlabs](https://community.topazlabs.com/u/kyle.topazlabs)\
**Post date:** [November 4, 2025, 6:54pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/94 "2025-11-04T18:54:29Z")

</div>

The model when run on a Mac requires 36GB of RAM to be used, trying to run the model with less RAM does not work for the process and the model to perform in the current version that it has been released in. The team is still working on it though and improvements could come in the future.

---

<div class="post-metadata">

**Author:** ![jo.vo](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/j/f19dbf/32.png) [@jo.vo](https://community.topazlabs.com/u/jo.vo)\
**Post date:** [November 4, 2025, 7:20pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/95 "2025-11-04T19:20:29Z")

</div>

> [@kyle.topazlabs](#):
>
> the model with less RAM does not work for the process and the model to perform in the current version that it has been released

Btw., did you acknowledge that your devs broke Starlight on Apple silicon in this V1.0.3?

It’s absolutely unusable in this state: quality is extremely bad - worse than the source most of the time and it only does1x no matter what output resolution/factor you select.

In V1.0 (as well as the last TVAI Beta) this did work. Why that extreme regression? When will this be fixed?

---

<div class="post-metadata">

**Author:** ![Splash](https://community.topazlabs.com/user_avatar/community.topazlabs.com/splash/32/230346_2.png) [@Splash](https://community.topazlabs.com/u/Splash)\
**Post date:** [November 4, 2025, 7:31pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/96 "2025-11-04T19:31:22Z")

</div>

Only problem with me so far is preset’s [doesn’t work] and the models loading. I downloaded the models I use but it still insists on downloading them and it takes 5 minutes to download and load models for it to do so. My only problems so far.

---

<div class="post-metadata">

**Author:** ![michaelandrews](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/m/df788c/32.png) [@michaelandrews](https://community.topazlabs.com/u/michaelandrews)\
**Post date:** [November 4, 2025, 10:19pm UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/97 "2025-11-04T22:19:48Z")

</div>

Anyone running one of these cards?  
NVD RTX PRO 6000  
NVIDIA RTX 6000 Ada  
if so what are your processing times?

---

<div class="post-metadata">

**Author:** ![david.counts](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/57b2e6/32.png) [@david.counts](https://community.topazlabs.com/u/david.counts)\
**Post date:** [November 5, 2025, 4:10am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/98 "2025-11-05T04:10:34Z")

</div>

I have Starlight Sharp and Starlight (Beta). Are you referring to “Starlight (Beta)” as “Starlight?” If so then why the (Beta) designation? Also, will M Macs ever be able to use codecs other than the default 8bit x263/x264 in mp4 wrappers when using Starlight Sharp? If not, then that makes it essentially useless in a broadcast environment which I work. Or am I missing something? Are Sharp cloud exports limited to 5 sec? I see no way to export a whole clip. Also, I’m seeing heavy GPU loads even with low power mode on. That doesn’t seem to redistribute any of the workload to CPU from GPU. Further, the audio is not copying over with Sharp. Starlight is exporting locally 1080 (original) when 2x is selected. In the export window it says Export 3840x2160 but is really 1080.

Topaz Video 1.0.3

M3 MacBook ProMax 128 Ram, 16/40 Cores exporting to internal.

---

<div class="post-metadata">

**Author:** ![david.counts](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/57b2e6/32.png) [@david.counts](https://community.topazlabs.com/u/david.counts)\
**Post date:** [November 5, 2025, 4:47am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/99 "2025-11-05T04:47:52Z")

</div>

Same. M3 MacBook ProMax

---

<div class="post-metadata">

**Author:** ![david.counts](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/57b2e6/32.png) [@david.counts](https://community.topazlabs.com/u/david.counts)\
**Post date:** [November 5, 2025, 5:05am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/100 "2025-11-05T05:05:04Z")

</div>

I”m getting x1 for x2, h264 instead of ProRes 422 HQ, mp4 wrapper instead of mov, export window showing incorrect resolution.

Source h264 1080 mp4 wrapper

M3 MacBook ProMaX 16/40 cores, 128 RAM

---

<div class="post-metadata">

**Author:** ![david.counts](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/d/57b2e6/32.png) [@david.counts](https://community.topazlabs.com/u/david.counts)\
**Post date:** [November 5, 2025, 5:08am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/101 "2025-11-05T05:08:01Z")

</div>

Good. I’m not crazy. Get the same results on a M3 MacBook Pro Max.

---

<div class="post-metadata">

**Author:** ![jo.vo](https://community.topazlabs.com/letter_avatar_proxy/v4/letter/j/f19dbf/32.png) [@jo.vo](https://community.topazlabs.com/u/jo.vo)\
**Post date:** [November 5, 2025, 7:01am UTC](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907/102 "2025-11-05T07:01:40Z")

</div>

Try the 1.0.0.  
There at least different upscale factors other than 1x work(ed). Not to speak about the overall massively better quality.  
I did not test container/codec capabilities, though as I normally export to h264 or h265 with high bitrate and then afterwards shrink the output with handbrake/software x265.

This is true for Starlight (formerly „mini“). Since SL sharp doesn’t work locally and imo isn’t really usable on the cloud, either with that silly 5 min limit I don’t use that. Besides I found SL sharp to be constantly inferior to SL mini for my content.

[Previous page](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907.md?page=4)

[Next page](https://community.topazlabs.com/t/topaz-video-1-0-3-patch/97907.md?page=6)
