For those without old manual lenses, yes, I think that might be the answer. You don't even have to shoot through it.
The end of 35mm adapers and hv20s....
- Thread starter Hunter Hampton
- Start date
For those without old manual lenses, yes, I think that might be the answer. You don't even have to shoot through it.
Matthew Bennett
Well-known member
I guess we're at ground zero, but it doesn't seem like anyone has been messing with different picture modes too much, like hi-con black and white and such, anyone even flipping picture-modes around at all to confirm that they do work?
I wish I had a car, I could drive to Buffalo right now and obtain a '90'
I wish I had a car, I could drive to Buffalo right now and obtain a '90'
His hands were pretty steady. It isn't like he has parkinson's.
Matthew Bennett
Well-known member
I've downloaded nearly all of the clip's original avi's from vimeo. Organics like flowers, bush, etc always looks really nice, or at least the median-blur thing seems to work with this imagery.
It's no big news that fine high-contrast lines at a distance are .. just plain annoying and terrible looking.
Hmm.. I guess its 'jungle nights' only for this...
Has anyone christened a name for the downsampling blur/median/watercolor filter yet? Something appropriately degrading that will stick in the Nikon rep's and engineer's heads, and in their next product version, it will say in the brochure, "AND NO MORE..{BLUR NAME HERE}"
I'll leave that up to the forum's poets..
It's no big news that fine high-contrast lines at a distance are .. just plain annoying and terrible looking.
Hmm.. I guess its 'jungle nights' only for this...
Has anyone christened a name for the downsampling blur/median/watercolor filter yet? Something appropriately degrading that will stick in the Nikon rep's and engineer's heads, and in their next product version, it will say in the brochure, "AND NO MORE..{BLUR NAME HERE}"
I'll leave that up to the forum's poets..
Can i ask which 28mm excactly? There are quite a few around. Is it a fully manual 28mm f/2.8 AIS?
For that matter has anyone tried a fully manual prime? Like Zeiss ZF or Nikkor AI or AIS lenses? I'm just wondering if they can be used since the D90 cant meter using those lenses.
Exporting clips into their individual frames, and then running a batch process in photoshop and importing them back into Premiere works beautifully.
Some of the filters in Photoshop do a much better job than Premiere. Not to mention being able to use filters that just aren't available for Premiere.
Anhar_Miah
Veteran
Steve I too do not wish to turn this into a long and protracted discussion and certainly not on this thread.
But please lets stay on factual grounds, and with all due respect making assumptions doesn’t help anyone:
In your post on page 222, you showed a correction algorithm that me and Barry had already discussed, I knew we discussed that particular paper because I recognised the photos.
All I did then was to merely point out on page 223 that such areas had already been discussed and that, corrections are not adequate, (please note, I did not at any point deny they didn’t work or not exist).
On page 224, you then incorrectly assumed that I was considering compensation on all scenes, even for the sake of argument if this was true, it still would not fix the problem presented by moving objects. In that same response you then listed some techniques, one was “3-D feature correspondences”, now sir with all due respect I know what feature correspondence is.
In my response I asked a specific a question regarding how 3d feature correspondence could even be used on a single camera on page 229:
INSTEAD of answering the above, you did not, I assumed you would present me with a paper for which this technique was used, instead you started “explaining” what feature correspondence is;
Now back then I let this slide, as I didn’t want to turn it into a longer discussion. I decided to inform you in a very polite and courteous manner that the explanation was not necessary;
Finally on page 234 you responded:
This is plainly false without any evidence to support such a false claim, why you would make such a statement I’m not entirely sure, I personally do not like to make assumptions of others.
In that same spirit, I will not make any assumptions why you made such a statement and leave it to the possibility that there was some misunderstanding or miscommunications.
Anhar,
But please lets stay on factual grounds, and with all due respect making assumptions doesn’t help anyone:
In your post on page 222, you showed a correction algorithm that me and Barry had already discussed, I knew we discussed that particular paper because I recognised the photos.
All I did then was to merely point out on page 223 that such areas had already been discussed and that, corrections are not adequate, (please note, I did not at any point deny they didn’t work or not exist).
On page 224, you then incorrectly assumed that I was considering compensation on all scenes, even for the sake of argument if this was true, it still would not fix the problem presented by moving objects. In that same response you then listed some techniques, one was “3-D feature correspondences”, now sir with all due respect I know what feature correspondence is.
In my response I asked a specific a question regarding how 3d feature correspondence could even be used on a single camera on page 229:
Anhar said:
INSTEAD of answering the above, you did not, I assumed you would present me with a paper for which this technique was used, instead you started “explaining” what feature correspondence is;
Now back then I let this slide, as I didn’t want to turn it into a longer discussion. I decided to inform you in a very polite and courteous manner that the explanation was not necessary;
Finally on page 234 you responded:
Steve Castle said:
This is plainly false without any evidence to support such a false claim, why you would make such a statement I’m not entirely sure, I personally do not like to make assumptions of others.
In that same spirit, I will not make any assumptions why you made such a statement and leave it to the possibility that there was some misunderstanding or miscommunications.
Anhar,
ilkka_nissila
New member
Full frame doesn't have "inherent vignetting problems", these are caused by lenses not optimized for it, basically the 70-200 and some superwides. If you use a lens optimized for FX, or most good 35mm film glass, no real vignetting problems.
Many DX lenses also have vignetting on DX cameras. E.g. 55-200 is quite bad.
Using DX cameras on lenses designed for 35mm film caused an illusion that vignetting is an uncommon thing. In reality it is an acceptable compromise in many cases that lens designers sometimes leave in to be able to correct other problems without making the lens too large.
The fact is, the same lens will vignette worse on FX digital than 35mm film, and that is the problem I was talking about. In general, today's lenses on FX wide open produce substantially more vignetting than DX lenses on DX do wide open. That's all I was really saying. This will change moving forward.
d90's mjpeg files are opened directly in cs3
since mjpeg is a series of jpegs if you output to frames or not won't affect performance
stoiqa
Well-known member
1.4 used at 2.0
On the posted settings,I know some of them don't affect vid,but since this thing is strange I just listed the current settings that I have on my cam.Even if Iso shows 3200,this will work only when the shutter button is pressed for pictures....will not work on vid mode.
Seems to me vid. mode has a life of it`s own.
D90 film Festival,who is up for it? :beer:
Last edited:
Steve Castle
Well-known member
Anhar, this is approaching silly, I merely pointed out a tool that could perhaps help some users here. I believe I have been nothing but polite with you and I've done my very best to be patient to explain to you from a technical perspective what I'm talking about. Its however hard to respond to vague statements like it "won't work" or its 'not adequate'.
I'll do my best to explain it one last time, the original algorithm you are discussing accounts for global and local movements, as long as the object doesn't fill the screen, rolling-shutter correction should not distort it. This algorithm first identifies the global movement of the frame and tries to estimate a velocity that can be used to make scanline realignments. The algorithm then goes in and identifies local movements and tries to get a better velocity estimation from the scene. Where moving object can be a problem is it can throw off the velocity estimation, however, the process takes priority to look at the entire frame globally rather then locally (this however doesn't help if the object fills most of the frame). How it is in practice you can test it out yourself with the program I first linked to.
But my point also is that the software side is moving forward rapidly, its surprising what can be done these days. As far as the use of 3-D feature correspondences in the use of rolling-shutter correction algorithms, in their newer articles they mention and reference other works that use the technique. My original point was that these techniques can be used in combination with each other to so that a more accurate velocity estimation can be achieved. If you like, I can PM you references and PDF files if you have further questions.
I'll do my best to explain it one last time, the original algorithm you are discussing accounts for global and local movements, as long as the object doesn't fill the screen, rolling-shutter correction should not distort it. This algorithm first identifies the global movement of the frame and tries to estimate a velocity that can be used to make scanline realignments. The algorithm then goes in and identifies local movements and tries to get a better velocity estimation from the scene. Where moving object can be a problem is it can throw off the velocity estimation, however, the process takes priority to look at the entire frame globally rather then locally (this however doesn't help if the object fills most of the frame). How it is in practice you can test it out yourself with the program I first linked to.
But my point also is that the software side is moving forward rapidly, its surprising what can be done these days. As far as the use of 3-D feature correspondences in the use of rolling-shutter correction algorithms, in their newer articles they mention and reference other works that use the technique. My original point was that these techniques can be used in combination with each other to so that a more accurate velocity estimation can be achieved. If you like, I can PM you references and PDF files if you have further questions.
Damn, you have a 28mm f/1.4?? Thats a very good lens and not cheap used.
stoiqa
Well-known member
el presidente
Well-known member
dont take my word for it but elswhere in this thread are many comments on very poor hdmi output(data overlays etc) only really any good for focus.
I think this thread is way too long now and there should be a seperate section for d90 camera, settings, lens etc.
I think this thread is way too long now and there should be a seperate section for d90 camera, settings, lens etc.
Anhar_Miah
Veteran
Steve, I've read the paper, I understand it, and its already been discussed, why you insist on "explaining" over the same point is beyond me, the simple question you never seem to answer perhaps is rooted in you not understanding the simple problem itself.
I feel I have been consistently polite, show me at any point if I have treated you otherwise.
Yes! please do PM me the references, THATS what I assumed you would do in the first instance instead of trying to "explain" something I already know.
Now please no more "explaining" if there is something factually incorrect, correct it, if I specifically ask what is "X,Y,Z" please ONLY then answer what is "X,Y,Z"
Sorry in advanced it this post sounds like I'm somewhat agitated.
EDIT::
Out of interest what is your background?
Last edited:
Steve Castle
Well-known member
I'm sorry, but quite frankly, nothing you have said previously suggests to me that you understand the problems that you presented. These algorithms do work with objects moving across the scene, even objects moving in the opposite direction (there are caveats and short-comings, which I have explained in detail). If you really did understand this you should respond to me with logical responses that address the technical issues at hand- not long rants.
I have given you technical explanation to all of your vague remarks and questions, its hard to respond to someone that says 'it won't work" when you do not give a technical explanation why. I feel I have had more patience with you then you have with the subject at hand.
With full awareness that this is a D90 thread, and consideration to others not to derail an otherwise interesting thread I will leave it at this (we can continue any discussion further via PM).
I will PM you the references and my background information as you requested.