Comparisons of ctv upscales vs mine

Close this topic

Hello! there are less artifacts on the ctv source. but not easy to compare because it’s not the same images on your screens

I think the darker one is underexposed and brighter one is overexposed. The correct exposure is somewhere in the middle, between the two. There are ways you can measure skin tone exposure using waveform or false color.

As you can see the gray indicator should be a standard for properly exposed skin tones. And true to what we see in your examples the skin is other over exposed or under exposed, usually leading to the rest of the scene being off the mark as well. The overexposed scene is slightly better, but not ideal in my view and according to the false color chart. In the chart, the gray area is where skin tones should be. There are different ideas where it should be, but I’ve found Flanders monitor to be closest to what I found works the best for my taste and eyes.


How do you adjust exposure, I would imagine its a matter of moving exposure slider in whatever NLE you are using, is it not?

I mean you could just load a video into video editor program, adjust exposure and export. Not much patience required.

I’m sorry, I won’t choose yours, there are a lot of artefacts and even traces of interlacing visible on the hand in the first photo… there is a slight lack of detail too even if yours is a a little sharper, it is better to take the CTV source and slightly increase the sharpness with an editing software to the limit and deepen it to correct the color, then you always have to add a slight grain when doing denoising to hide the effect “plastic” and the few minimal artifacts that may remain. Last night I found a new method to lose even less detail, is to use vegas pro to deinterlace, since it produces artifacts, is to then import the deinterlaced video with hybrid, and with QTGMC, put “input type” on “2” or “3”, it fixes artifacts, and play with “keep grain”, then produce the video with proteus in vai. I have slight loss of details when there is movement even with “final temporal smoothing” on “0” with QTGMC but probably because I put the compression at 20 on proteus. I will still refine that, but it is by proceeding like that that the details are the most realistic, once the pass with proteus is finished, I refine the settings with vegas pro, that gives me this:

it’s a source of 8 mm cassette, so hard enough to improve. (enhancement is in hd)

1 Like

are you telling me that you are making an improvement without deinterlacing your source? loool and it’s me who don’t know what I’m doing? I’ll give you my opinion, I don’t think you’re experienced yet, look:

where it is circled is traces of interlacing for me, how can your source be cleaner than the other if it is not deinterlaced? the red circle at the very top, I notice that there are less details than on the CTV source.

(20 compression in my case does not give me a plastic effect), and you always have to add a minimum of grain in the end;)

put the same images in comparison already to see better, or send me a sample video, I may give you a better opinion

the fifth word of your second sentence for me :wink:
and i understood that you didn’t deinterlace… i died laughing! when do you enhance a video on an interlaced source? deinterlace it first go! I don’t see your source any clearer, but less beautiful in any case x) the goal of an improvement is to remove any interlacing and any artefacts. do that first, with a minimal loss of detail and then you manage the sharpness… it’s the basis in the improvement in fact. so don’t be bitter with me when you’re working on an interlaced source. :')
you will see with experience that the fact of deinterlacing a source well will make you lose details, something that you must manage to recover, and it is not easy, believe me :slight_smile:

now I gave you my advice, do what you want

no problem for me, i have other things to do better… only donkeys don’t change their minds haha
it was you who got carried away when I gave you my opinion… and things that are not true you say… so you are working on an INTERLACED source, from this moment, I see that you don’t know what you are doing…

1 Like

you delete your messages because you realize your mistake?

Have you already tried to deinterlace only with QTGMC but keep it simple?
No denoising only deinterlacing using the preset Very Slow in Hybrid/Avisynth.
In my opinion it does an excellent job in denoising and artifact removal without destroying details.
About Proteus… In my opinion v2 it’s much better than v3. It’s not so artificial

I tried everything with qtgmc, when you don’t touch “final temporal smoothing” there is almost no loss of detail in the movements. I use qtgmc with vapoursynthe in hybrid. and when you set a preset to slow or veryslow, it deinterlaces very well but there is a loss of detail, when you set fast or veryfast, the details are preserved but it deinterlaces badly. so I found out that deinterlacing with vegas + qtgmc with the input type set to 1 or 2 gives me much better results. i’m not experienced enough with proteus, not long ago i was using artemis hight and medium, which can produce better results than proteus depending on the noise of the source. now i am using proteus with version 3.1.8. but someone said recently that this one produces better results than older versions. so I don’t really know, I just noticed that denoising is better managed than on version 2.6.4

1 Like

in fact there is grain and noise, it’s different, when you add noise, it’s before the treatment, the more you put in and the more details you’ll lose, the grain at the bottom, it’s is applied after the treatment, and that’s what you have to put on to hide imperfections, or not to see a plastic effect. I do like this for a 720 x 576 dv source: amount 2 size 1. personally I don’t have to put any because I apply other filters with vegas pro, but for a final video, it’s better to send some and it has always been recommended. for the lines of your video, you won’t necessarily see them when there is no movement, but I advise you to deinterlace your video before making any improvements, it’s not even advice in fact it’s an obligation … you should try the dione dv and dione td models, but it’s not the best, right now qtgmc is the best for deinterlacing any source.
ps: at first I did not argue anything, but simply gave my opinion on your comparisons. so it’s true that it’s hard to see because I notice that the quality of the photos sent here is limited. but we clearly see lines of interlacing on your first photos, something that nobody can conceive in fact…

What do you mean with “when you set fast or veryfast, the details are preserved but it deinterlaces badly” ? Leaves visible interlaced lines is that it?

yes basically the slow presets deinterlace well with straight lines but in fast, the lines have some artefacts, on some passages in fact, to see in this photo the edge of the boat on the right, we see that with a slow preset, we still loses detail.

what I don’t understand is that there are several settings, preset, source matching, noise preset, match preset, and match preset 2, I don’t master everything yet. and before I deinterlaced with vegas pro, or interlaced with vegas pro to send to hybrid because my reference video for the test is 1h20 and it never loaded with hybrid, it’s because the video was on a classic hard drive, but I moved it to an ssd, and it loads normally but I still have to wait a good 3 minutes. in the end i can do direct deinterlacing with hybrid and my original source :slight_smile:

Sorry for the delay in replying…
I can see what you mean. It’s not easy to deinterlace a source and keep all the details but as long as the most important are preserved…
I think you are like me, a perfectionist :wink:
Do you know which codec Vegas Pro used for output?
It seems to me that something like Uncompressed RGB was used, that is why you have problems loading it fast enough in Hybrid or even playing it.

Exactly, I don’t really want to lose any details on my family videos, that’s why I struggle so much otherwise I will have finished everything for a few years already haha with vegas pro I always go out in uncompressed avi, in fact my one-hour video did not load with hybrid unless I put it on an SSD, But the same shorter videos load even if they are on a regular hard drive. The videos are only 22000/23000 kb/s though.