[brazen1]

DJ. This is the 2nd time you've jumped all over me and why I don't know? I've seen you around. I've always had respect for you. If I'm not mistaken, you are in the video editing business. That's why I respect you. Perhaps you could teach us and set us straight when we are not "serious", "trolling", or the likes, me in particular.

I'm no expert like yourself but, this is my understanding......
3D for example. The best you are ever going to get is half resolution for each eye. I'd like to watch my movie at full resolution for each eye. So, my combined resolution has to be doubled to pull it off. That's my logic. I don't think Kodi can do that? I set my upscale at 100%. During file play, I check the audio/video/codec whatever it's called window and it shows 1920x1080 before I software upscale a title. The original Blu-ray. So, I assume Kodi isn't doing anything. After I encode it, I check again and it shows 3840x2160. To my eye and the eyes of others when we compare the 2, the larger one stands out.

Now maybe we just want to think it's better when it actually isn't except we aren't the type. Someone told me a saying long ago. "Some people listen to their music. Some people listen to their music system." We are not the latter.

So, yes I am serious, I'm not trolling and this isn't a satire. Perhaps you'd be kind enough to educate me slightly why I'm not and where I am wrong and why? Your last dig at me on another thread left me as perplexed as I am now. Fwiw, I still respect what you do, just not how you do it.

[DJ_Izumi]

I'm gonna be totally honest with you here. Unless you're using some kind if interpolation when upscaling Kodi and whatever re-encoding program you're no real gain. Even with an interpolation method you really have a computer trying to GUESS what new pixels should exist based on a comparison of surrounding pixels. But still, all you have is a GUESS so such systems are generally not that pleasing in all cases, though animation often benefits more easily. I am curious what solution you're actually using for scaling things up.

I honestly think that you are experiencing the placebo effect, you think you are doing something improve things, which creates a cognitive bias for you so when faced with two examples you 'feel' the one you 'think' has been improved looks better even though there's no real difference. I think you're wasting processing time and hard drive space to make something bigger without actually improving it.

[mellomade]

AND THERE ARE VISIBLE DIFFERENCES. (and measurable too, I'm no audiophile)..

Really, as storage, processing power, and bandwidth continue their Mores Law march, the need for compression will decline.

and the unfortunate soul that that chooses 3 GB files rather than 30 GB files will be left with a xerox of a xerox of a xerox of a xerox..

File size has to do with image compression but also (and more importantly) bit rate variability. Just because a disc is encoded for 40 MBit/s doesn't mean all the data requires this. Most of the time it does not - and FFMPEG can deal with encoding variable bit rates quite well which can significantly reduce file size while maintaining quality.

Of course there will be differences - this is lossy compression after all. The question rather is of perceivable difference. If I have to stand 2 inches from my display and pixel peep while straining my eyes to see the blocking and other artifacts this to me is acceptable loss. The point is that when properly done - an encode that takes up 6Gb vs 30Gb with little percievable difference is certainly worth IMHO the thousands of dollars I am saving on server hardware.

And that is a really funny part of this conversation. Usually in my experience the folks contributing to the Hardware section here on the forums are quite cheap. I don't mean that in a derogitory way - it's just that there is a LOT of bitching about $150 playback devices and those here don't bat an eye at the thousands of dollars in server hardware that raw MKV dumps require. The irony is amusing.

If that is so, the rhetorucal question is: why are blurays produced at all? The 9G available on a DVD would be enough.

haha - then what would the studios do with all the crapware they put on these discs?

[brazen1]

Thanks for the answer DJ. Yes, using interpolation that has no choice but to guess how to anticipate new pixels. About the biased opinion, it boils down to this:

Using Kodi, I can't play a native 3D. An external player can't either. It's because my panel only offers (3) 3D modes, none of which can handle a native 3D disc.
I can 'trick' my panel to add an extra mode. Checkerboard. That trick kills display settings reverting to limp mode that looks terrible so I'm stuck with SBS or TB.

I have no choice but to put the movie into a container that Kodi and my panel are compatible with. An MKV. That means I have to strip the Blu-ray and convert it. I figure as long as I have to convert it I might as well upscale it while I'm at it. I'm using a good ol' retail software to do it. I have no access to commercial equipment nor the knowledge to use it if I did. Yes, the conversion takes 10 times longer since I'm upscaling and stressing my CPU. I Don't care. That's why I have a tool to do the job and I'm in no hurry. The extra GB's it adds, I don't care. I'd imagine if a popular retail software to upscale was a farce and wasn't doing anything but wasting enthusiasts time and money, it wouldn't have been developed, but what do I know about large companies ripping off the public?

It's difficult to watch a scene from the original rip, stop it, load the 'improved' rip and watch the same scene and pick out differences. But we have. We even tried to do the placebo thing. Telling the watchers they are viewing the new and improved rip when it was actually the original. We got proper feedback. To make things even more difficult, we pause a frame, take a snapshot and do the same for the 'improved' rip on the same frame and compare the frames side by side. Things get really interesting because each snapshot is actually snapshotting (2) SBS frames and not what you see through the glasses. So when comparing, we have 4 snapshots of frames to review. The few I've done, I saw subtle differences. Was it worth it? I think so and if we are all fooling ourselves, well no huge harm done because the 'improved' rips certainly are not worse than the original.

As for shrinking my thoughts are this and I could be wrong. If you shrink, you are going to lose quality. If you view that shrink on a smart phone, you won't notice a thing and you should shrink. However, if you are going to view on a 65" panel or even bigger projection, you might want to keep as much quality as you can. Another thought. What looks acceptable on your shrink today might not look so good when you upgrade to bigger, more hi def panel in the future.

My intentions are to wait on h.265 hardware and rip the entire collection again. Hopefully smaller this time. It will probably take a year. As I mentioned in the other thread, I already tried h.265. Panning scenes were out of the question. My CPU was not stressed. My audio was in sync. Everything worked well..... except panning scenes. It skipped frames. Knowing it wasn't my CPU not keeping up, I have no answer why this occurred? I actually tried with 2 different CPU's. Maybe GPU hardware will cure that, maybe not?

Thanks again for your opinion and being nice. We be cool : )

[DJ_Izumi]

I'd imagine if a popular retail software to upscale was a farce and wasn't doing anything but wasting enthusiasts time and money, it wouldn't have been developed, but what do I know about large companies ripping off the public?

I'll be frank, Kodi will use the same methods to scale 1080p video to fit it's 4K scale as your program does, and it will do it in real time. It's the same thing that happens when I play standard definition video on my Kodi system hooked up to a 1080p display, it scales that 480p content up to fit a 1080p frame. It's honestly just SIILLY that you are upscaling in a re-encode just to 'Make it 4K'.

It's really that resizing options in that kind of software has OTHER purposes. Some devices only accept very specific sets of resolutions and content must be resized when making a video compatible with those devices. Maybe someone is converting a mix of SD and HD content into a single Blu-Ray video disc, where everything has to be uniform. What it ISN'T there is to magically make your video 4K in an effort to 'improve it'. You are honestly just demonstrating a silly, comical naivety about video.

Lemme put it this way, let's think about this as color instead of resolution. If I take a black and white image, that is stored only with grey scale colors, and I convert it to an image format that supports millions of colors, does the image MAGICALLY become a color? No, it's just a greyscale image now in a format that is using a LOT more data for color, all of which says 'Nope, No color here'.

[skylarking]

I've pondered this upscaled re-encode issue previously but for me it comes down to not having the required knowledge to really know what the encoders are doing verses the capability of the hardware in real time.
As is, my Pioneer TV has what many would consider resonably good quality upscaling circirty and/or processing grunt, like my PS3, so i wouldn't benifit much by upscaling re-encode of the source before playback.
But if the TV has poor quality upscalers or the kodi device doesn't have the hardware capabilities or grunt to upscale well, in real time, then pre-processing the source may be an option to real time processing.

And i've seen some rather cheap HD TV's with very poor PQ when viewing SD but OK PQ when viewing HD.
In this context, a de-interlace and upscale re-encode of SD video when some low power gruntless box running kodi will play back on a cheap 1080 TV may yield better results than real time processing on the kodi box or cheap tv
Just saying...

[DJ_Izumi]

And i've seen some rather cheap HD TV's with very poor PQ when viewing SD but OK PQ when viewing HD.
In this context, a de-interlace and upscale re-encode of SD video when some low power gruntless box running kodi will play back on a cheap 1080 TV may yield better results than real time processing on the kodi box or cheap tv
Just saying...

Deintercing sure, deinterlacing, when done exceptionally well, especially if it's an inverse-telecine, can be difficult to do in real time at best, and re-encoding to progressive can get you lots of benefits, but the upscale is still a waste of time. When I rip my DVDs to HEVC, I obviously deinterlace or inverse-telecine them, but I still leave them at their native resolution. ...That said, NLMeans noise reduction is a pretty awesome way to lean up DVDs made from lower quality masters. MONSTERIOUSLY demanding in processing though, like 6fps on an Intel quad hex core but the results are amazing.

[kyle84]

Of course there will be differences - this is lossy compression after all. The question rather is of perceivable difference. If I have to stand 2 inches from my display and pixel peep while straining my eyes to see the blocking and other artifacts this to me is acceptable loss. The point is that when properly done - an encode that takes up 6Gb vs 30Gb with little percievable difference is certainly worth IMHO the thousands of dollars I am saving on server hardware.

And that is a really funny part of this conversation. Usually in my experience the folks contributing to the Hardware section here on the forums are quite cheap. I don't mean that in a derogitory way - it's just that there is a LOT of bitching about $150 playback devices and those here don't bat an eye at the thousands of dollars in server hardware that raw MKV dumps require. The irony is amusing.

I don't think it's accurate at all to suggest that raw blu-ray rips require thousands of dollars worth of server hardware. While a full rip does require more storage space, what makes things expensive are the redundancy mechanisms you choose to implement for your data. Which, depending on your needs, may or may not be necessary.

[brazen1]

So, in essence, what you are saying is you can't squeeze blood from a turnip and I am basically putting lipstick on a pig? Before I started 'improving' my rips by upscaling, my logic tried to tell me how can I possibly get more quality than what I started with? I figured if the media properties showed the resolution doubled, then it must have, especially knowing the file size grew and that must mean there is more information added. But you're saying that isn't the case and I'm beginning to believe you. Perhaps the placebo effect is influencing more than I think? I can't get around one thing though and maybe you can set me straight? Since the file size grew (and remember, extra video, audio, and subs were stripped) is that additional added information worthless? I assume you're trying to convey those extra added guessed pixels are most likely wrong guesses especially if they are not guessing animation?

Thank you for interacting. I was hoping to finally land on a forum/thread to educate my confusion and assumptions. I'm naïve to all of this and interested in learning more. Not to get too far off topic from this posters thread, I have other questions related to compressing and I value the input coming in from everyone. Rather then start a new thread, I think this is the proper place?

[brazen1]

Since I was upscaling h.265, I now believe that is why I had skipped frames during panning scenes. I'd like to convert h.264 to h.265 so I can take advantage of compressing the file without losing quality where it would have using h.264. Here are a few questions that come to mind:

I've never been fond of default values in the ripping softwares. Selecting 'High Quality' and the resulting file goes from 42GB to 5GB with plenty of artifacting, dull colors, and fuzzy picture is not my idea of acceptable 'High Quality'. I understand acceptable quality to your eye may be different than mine. In your expertise, knowing I'm seeking the highest retention in quality vs. file size, how would you start to adjust settings for DvdFab (if you are familiar with it)? If you suggest switching to a different software like handbrake or something and it's settings, that's not a problem. Every thread I read suggests to just rip with default settings, watch it and see if you like it. I'm not trying to cut corners. The default looks terrible. Rather than experiment with zillions of settings, I'm hoping you guys already have. What are the maximum settings without going beyond perception of the human eye? If I set to that, I know it isn't going to get any better, I've not exceeded the settings beyond what I could actually see, and leave it at that. If there is a little overkill, that's ok too.

These are the only relevant settings DvdFab offers:

(1) Encoding Method. 1 pass, 2 pass or CRF? I understand 2 pass 1st analyzes so it can anticipate (guess) the pixel layout. Is it necessary and is the end result actually higher quality?

(2) Bitrate. I've always maxed it to 0.5. Is this well beyond actual perception and should it be reasonably lowered without losing quality now that it will be h.265? This has everything to do with final file size. For instance, would lowering it to .30 bring it down to that fine line of actual perception without crossing the line of actually seeing quality loss?

(3) Is it safe to assume every title would have drastic different settings from each other or would the settings basically be the same for each movie?

Thanks for any opinions you might offer.

[mellomade]

I don't think it's accurate at all to suggest that raw blu-ray rips require thousands of dollars worth of server hardware.

How do you figure? If you have 500 movies (which I would think is on the lower end of what folks have here) at 25Gb per movie you need roughly 25Tb of storage including backup. If a single 3Tb drive costs $130US - you are already at $1000 for the drives alone.

While a full rip does require more storage space, what makes things expensive are the redundancy mechanisms you choose to implement for your data. Which, depending on your needs, may or may not be necessary.

Roughly 6x the storage space! I've spent the last 8 years curating my collection of 500 movies. No backup or redundancy? That would be just plain irresponsible. No way I want to re-rip my collection if/when something happens to the drives on my main server. The time you save not compressing raw rips will certainly evaporate if you don't back your stuff up!

[mellomade]

@brazen1 - FWIW - here are my video settings for Handbrake for BluRay discs. For faster action/low light films I might tweak the RF value down to 18. Otherwise RF20 covers most discs quite well.

[nickr]

Brazen, what is your screen's resolution?

[DJ_Izumi]

I can't get around one thing though and maybe you can set me straight? Since the file size grew (and remember, extra video, audio, and subs were stripped) is that additional added information worthless? I assume you're trying to convey those extra added guessed pixels are most likely wrong guesses especially if they are not guessing animation?

To be clear, animation would still require an advanced interpolation method, which would still only work really nicely on cell animation, not really CGI or really detailed/textured 2D animation that some houses produce. But... Yes. You are creating MORE information and your file is larger because of this, the problem is it's just making a blurry 'guess' between the pixels as it stretches the image out but it now has to store all the data, including the blurry stretched out image and it will take more storage for that. The thing is, you haven't GAINED anything other than file size, you haven't gained any more INFORMATION, in terms of detail, than you had before. How about I try to explain this visually.

This is a 100x100 JPEG image. It is 34.4KB but 100x100 pixels is not very high resolution.




Let's stretch this out to make it bigger! Stretched out to 800x800. But now it's blurry. All we did was stretch the existing information out into a larger space. This doesn't give it any new DETAIL however so it's a blurry mess. We did however gain file size, it's 251KB, but while more storage is needed to hold all this stretched out information, it gives no DETAIL. Also, Kodi could do this stretch on the original 100x100 image during playback instead, so we wouldn't need this much larger file. All we've done is waste storage.




Now, for comparison, this is what the image would look like if instead of being resized from a 100x100 original, but from a very high resolution original, like that which movies are filmed with. Most movies these days are filmed at resolution well BEYOND 1080p and then it's scaled DOWN to fit on a 1080p bluray. Look at all this detail. This is all the detail we did NOT gain just by stretching up a 100x100 version. This is the detail you'd see once 4K BluRays or other 4K sources come, but NOT just by stretching out lower resolution material to 4K.




On a side note, my new arcade sticks for my Kodi/Steam Machine are awesome. :3

[robo989]

The thing is Robo, I can taste things that many others cannot. I assume that some people can perceive things that I cannot. It could be they imagine it too. I don't know.

The point is none of us have a dog in the hunt how someone else rips their library. In the end the only thing that matters is we are all happy with our methods.

A good common sense answer, but not really true
It is subjective, to a point - but not subjective in the way that art is subjective for example, that is to say that an obvious degradation in quality will be noticeable to all, as CRF value is lowered, so does the number of people that can see the difference.
PSNR can be measured with an absolute, though it's hardly a modern way of A:B comparison's it can be measured, it is not art.
x264/x265 using CRF mode can give an absolute quality level, that is to say I would challenge anyone to tell the difference between a CRF16 encode of a 1080p source that has no grain to the original and see a difference.
They won't find a difference.

"fast moving scenes" haven't been a problem for years, in response to a another poster. CRF encodes with a consistent quality, with bitrate increased to maintain a constant quality, not a 2 pass guesswork.
Of course you can't predict file size.


The real advancement has been with audio anyway last few years that's actually useable now, have just finished doing about a 1000 conversions from DTSMA to Opus...3TB/~20% saving.