MP 1,1-5,1 Topaz Video AI On cMP

[Dolphins1972]

Curious - what settings in Topaz made it worse for you? The results I get seem pretty good except for the file size.

I found a way to reduce the file size by as much as several gigs without reducing resolution or any loss of quality. After upscaling, re-encode with HandBrake, choose the settings that works best for you. Re-encoding can reduce the file size by more than half depending on the settings you choose.

 

[phrehdd]

I found a way to reduce the file size by as much as several gigs without reducing resolution or any loss of quality. After upscaling, re-encode with HandBrake, choose the settings that works best for you. Re-encoding can reduce the file size by more than half depending on the settings you choose.

First, thanks for bringing up other software such as Handbrake and how it may reduce the size of files and has many options to choose from as well as settings. Second, we got to be really careful when talking about measures of such files.

When a service streams to you a 4k movie, it is not the size of the file(s) that make up a 4k movie on a disc you buy of a movie that is 4k (UHD). So how can two 4k offering be so different in size? Answer - some detail is lost and, often th compression used may have tricks to avoid redundant information such as areas of an image that remain static. The truth is the streaming version is not as good as the disc. This is not my opinion but a fact.

Handbrake more often than not is reducing a file which is in fact "lossee" meaning data has been lost from the original that it came from. While it may be possible to compress some data slightly (media files), you are already coming from a compressed file on the disc. Compresson on top of compression is in fact, missing info/detail found in the original disc.

Perhaps a better way to state things is that one might be able to compress a movie file, have the same resolution and be visually acceptable or no perceived loss of detail.

 

[Melbourne Park]

Yeah, the only way I'll see better results is with the Pixlas Mod and a newer GPU.

I've now gone a new MacBook (currently going to test a M3 40 core Max refurb, have until Jan 8th for a refund). Not sure it will be as quick as an M4 Pro (rather than the much more costly M4 Pro max).

But no need for Pixla's mod as you can always put a power supply outside and just put the cables through. Not as aesthetically pleasing but now the 5,1 is pretty worthless. I looked hard at 7,1s and realised the 5,1 has almost the same short life left with Mac OS as the 7,1, although the 7,1 can run the current OS properly unlike the 5,1. I'd still love a 7,1 but unfortunately the game has changed ... with Apple, anyhow. Been through quite a few Apple desktops with slots inside and still have a couple, but most of time time is spent in front of a MacBook anyhow ... I feel some loss not ever having had a 7,1.

 

[Dolphins1972]

I've now gone a new MacBook (currently going to test a M3 40 core Max refurb, have until Jan 8th for a refund). Not sure it will be as quick as an M4 Pro (rather than the much more costly M4 Pro).

But no need for Pixla's mod as you can always put a power supply one the top and just put the cables through. Not as aesthetically pleasing but now the 5,1 is pretty worthless. I looked hard at 7,1s and realised the 5,1 has almost the same short life left with Mac OS as the 7,1, although the 7,1 can run the current OS properly unlike the 5,1. I'd still love a 7,1 but unfortunately the game has changed ... with Apple, anyhow. Been through quite a few Apple desktops with slots inside and still have a couple, but most of time time is spent in front of a MacBook anyhow ... I feel some loss not ever had a 7,1.

I hear ya. My dreams of owning a 2023 7,1 and taking it for long walks down the upgrade aisles on glorified $700.00 skateboard wheels were dashed to pieces by Apple's decision to make them obsolete in a years time.

 

[Dolphins1972]

First, thanks for bringing up other software such as Handbrake and how it may reduce the size of files and has many options to choose from as well as settings. Second, we got to be really careful when talking about measures of such files.

When a service streams to you a 4k movie, it is not the size of the file(s) that make up a 4k movie on a disc you buy of a movie that is 4k (UHD). So how can two 4k offering be so different in size? Answer - some detail is lost and, often th compression used may have tricks to avoid redundant information such as areas of an image that remain static. The truth is the streaming version is not as good as the disc. This is not my opinion but a fact.

Handbrake more often than not is reducing a file which is in fact "lossee" meaning data has been lost from the original that it came from. While it may be possible to compress some data slightly (media files), you are already coming from a compressed file on the disc. Compresson on top of compression is in fact, missing info/detail found in the original disc.

Perhaps a better way to state things is that one might be able to compress a movie file, have the same resolution and be visually acceptable or no perceived loss of detail.

I won't argue any of that but unless you're viewing the media on an 8K, 60" UHD monitor, I doubt any loss of quality would be perceivable to the human eye? I have a Samsung 30" 4K monitor and I can't tell the difference between the Topaz upscaled and the HandBrake re-encoded files. It depends on the settings you use to re-encode. HandBrake does have lossless settings but using them won't reduce the file size.

 

[phrehdd]

I won't argue any of that but unless you're viewing the media on an 8K, 60" UHD monitor, I doubt any loss of quality would be perceivable to the human eye? I have a Samsung 30" 4K monitor and I can't tell the difference between the Topaz upscaled and the HandBrake re-encoded files. It depends on the settings you use to re-encode. HandBrake does have lossless settings but using them won't reduce the file size.

On 32" monitor and 55" Large screen TV - handbrake compressed 4k files do show at times crushed blacks, blocked out highlights, fast moving objects not as clear etc. What is clear however, is that some people won't mind or notice unless they have a reference point (comparison). In short, if it doesn't bother you to have this quality of image for a smaller file then have at it. The only time I used Handbrake was to reduce resolution for other formats (tablets and phones as example).

The struggle to get the best image quality when further compressing an image remains as does in some instances, upscaling for 4k TV. I have found that for a 4k file, NVidia TV Shield (pro) does an excellent job overall for 1080p and similar up to UHD/4k. It beats out other streamer devices and some receivers (AVR) and the television itself. As I say the latter, many people are happy with the way their tv upscale 1080p files etc. I tend to want to be a match or as close to the original file media 'goodness.'

 

[Melbourne Park]

Funny about the Shield! Fact is my LG G3 OLED does enhance old movies. But I want the resolution to be available to any viewing platform. Rubbish in rubbish out is my theory.

Topaz for my task used I think 29 GB of RAM. For 720x576 resolution to 4k at 25 fps, the MacBook Pro M3 Max 40 core with 64 GB RAM and a 1 TB drive that BlackMagic tests at 6450 write and 4800 Read, I was getting from about 6.6 frames per second. From 4 up to 7. Overall I think it was 6.1 fps. With a 20% sharpening enhancement option selected. Very good results although the yellows look a bit too rich. Viewed on a 75" G3 LG OLED TV streamed direct to the TV Google Nest Pro which supports I think from the MacBook via WiFi (the RV's WiFi is ethernet, the Wifi device is a Google Nest Pro which supports 6E). This was before editing in FCP. I am just testing the computer. Do brewing on the net did not seem to effect performance of Topaz or the browser. I reckon that for my amateur usage, I might be better of with a M4 Pro although 48 GB of RAM annoys me. But it appeared on the first tests I did, the the Topaz was only using 28 GB of RAM.

CPU usage did not seem an issue, which makes me think that an M4 Pro might be faster? However GPU usage seemed to be significant, and that means more cores should assist?

I did some screen captures too, memory and also CPU. Did not do Disk though. :

But despite the 64 GB RAM there was 27 GB cached files.
I need to test:
- after restarting,
- use Topaz
also
- the screen capture but not save screen capture to photos as doing so opens up Photos which uses RAM.
- Activity monitor

Run nothing else while do such tests.

 

[Melbourne Park]

I hear ya. My dreams of owning a 2023 7,1 and taking it for long walks down the upgrade aisles on glorified $700.00 skateboard wheels were dashed to pieces by Apple's decision to make them obsolete in a years time. View attachment 2451580

Me too with my desk chair wheels (matches the red alloy and leather chair) but I don't think chrome was ever available, I thought Apple cornered that market?

Heck I need to clean the floor, been moving a lot of house junk around ...