fuzzy resolution from files imported from Photoshop

I?m an artist and I?ve been making video?s that are comprised totally of stills I make in Photoshop from modified photographs. Each frame is slightly different, so that importing them in order gives the feeling of movement, much like a kid?s flip book. I know there are much easier ways to do the same thing, motion keyframes, After Effects, whatever, but I love the jerky quality I get so I don?t mind the enormous amount of work involved.

My problem comes with resolution. My original Photoshop files look great. I have to make videos that look as good as possible whether in DVD format for projecting on a wall in a gallery or museum, for playing on a computer (a Quick time movie) and for standard DVD playback on any screen.

My images are quite complex and there is a lot of movement. I can?t find a resolution to make my stills in Photoshop that ends up with a crisp image in the final output. They are always fairly fuzzy looking, I export the files with uncompressed 10 bit for Quick Time and I have yet to be able to burn a DVD with good quality. I have been experimenting with a variety of settings Whether I use 720 x 486, 720 x 547, 720 x 547 , 1280 x 720 or 1920 x 1080 the only differences seem to be whether or not the image looks stretched, not the sharpness of the image, otherwise the results look pretty much the same. I have been using ?square pixels?.

Could you please recommend what are the optimal settings for both the initial Photoshop files, as well as how to compress with the least compression possible when exporting.

what color space are you using in photoshop? CYMK or RGB
photoshop must be set to rgb to be converted to video ithink. i am in a class now for photo shop and we learned this yesterday.

in fcp does the photos look blurred? if the photo is clear in fcp then it made it there ok and your trouble shooting is in the fcp export which there are a lot of things that you can do to get a clear export.

as far as the dvd the problem there is the mpeg2 codec it always down grades the clarity. you could however atemp to burn hdv and that will help it look better i would thing. i have never burned an hdv disk but i have upres dv from interlace 30i to hdv 30p. this will change the over all texture of the movie but in most cases it looks just fine.

""" What you do with what you have, is more important than what you could do, with what you don't have."

> > > Knowledge + Action = Wisdom - J. Corbett 1992 """"

Making PSD images look good in Final Cut is a bit of an art in itself.

There are several factors involved.

I would start with the obvious.
As stated by J Corbett make sure the files are in RGB.
In terms of sizing and resolution I would make sure that the file is as big in pixels as the maximum size you will be displaying.
So if you are working in PAL simply save the images as 720 X 576 pixels and 300dpi.
I photoshop there is a canvass preset for PAL or NTSC with aspect ratios etc already in place. Use one of them but increase the resolution.
If I undestand you correctly you are just displaying the images without actually animating their size or position in Final Cut, so there is no need to save the images any bigger.

In terms of exporting the QT there is no point exporting it in a different codec than the sequence setting you are using if you want to maintain a high quality file.
The DV codec should be fine but I would make sure it is not interlaced.

Its basically all about keeping the files in the same size and resolution throughout the prcess and onto the final DVD.

Johan Polhem
Motion Graphics
www.johanpolhem.com

I'm sorry, I'm new to FCP, it amazes me the number of posts I've read about this sort of issue.

I simply created some graphics in Motion imported them into FCP, they looked great in the pre-view window but as soon as I placed them into the timeline they appeared soft in the program window and were indeed soft when I exported them.

I tried different codecs, incompressed, animation etc.. same result.

While doing a green scree test I noticed I had the asme problem on the green screen footage I was importing, at this point I was erady to trash FCP. However for some reason I knew this had to be a filtering problem with FCP, there was no way that I was getting the same result from so many different sources of source material so I did what any self respecting editor would do.

In the same project that you are currently working, open a new sequence, make sure the sequence settings match, cut and paste from your current sequence to the one you just created and see if this solves your problem.

I don't see why this should work but it worked for the three projects that I was having this problem with. Besides its easy and quick to find out.

I'm using FCP 5.1.2 on a Mac Pro 10.4.8.

What are your sequence settings? What is the codec of the footage you are adding to the sequence? Animation codec will look GREAT in the Viewer, as the viewer shows you what the footage looks like. But, if you drop it into a DV timeline, your footage is instantly compressed 5:1 in the DV codec...so it won't look nearly as good.

BUT...you simply cannot judge the quality of anything using the computer monitor. Never ever ever. Only judge the quality of your image on an external monitor connected to the system.

If it still looks not as good, well, the source image is obviously better quality than the codec you are editing with. This is very common.


www.shanerosseditor.com

Listen to THE EDIT BAY Podcast on iTunes
[itunes.apple.com]

Thanks for the many responses.
I am using RGB in Photoshop.
I have tried numerous tests at a variety of resolutions. Since I am working in NTSC, I have tried everything from 720x547 all the way to 1920x1080. I am using Uncompressed 10 bit project setting and I'm exporting to Quicktime Movie at the same setting. Thank you Ken Stone for suggesting that, which has helped a lot. (I have yet to attempt the DVD).

To:Johan Polham...if you use an NTSC preset sets the pixel size, which means the resolution will only increase by decreasing the file measurements, the size of the file stays the same. This means NTSC at 72 is exactly the same size file as NTSC 300. Increasing the resolution in Photoshop therefore won't help. But thanks for the reminder not to interlace. (I don't think I did, but I'll certainly check next time.)

To: Shane Ross...There is no codec of any footage I am adding to the sequence--since they are all stills of photoshop files, there is no footage. The appearance on the monitor is not important to me, just the quality of the final DVD...but if the Quicktime movie is soft, the DVD is not going to get any better.

CB, are you judging the quality on your mac monitor from FCP or are you looking at this on an external NTSC monitor?

Mac monitor....also on a wall from a projector, which is how it will often be viewed. Eventually I will need a DVD, which I haven't burned yet since the Quicktime movie isn't wonderful.

Did you try creating a new sequence and cutting and pasting from the old to the new?

"BUT...you simply cannot judge the quality of anything using the computer monitor. Never ever ever. Only judge the quality of your image on an external monitor connected to the system."

> but if the Quicktime movie is soft, the DVD is not going to get any better.

Not necessarily true. If you're using QuickTime versions before QuickTime 7, QuickTime movies exported from FCP by default have "High Quality" disabled. Open your movie file in QuickTime Player and press APPLE-J. Navigate to your Video Track and make sure "High Quality" is enabled. If it used to be off, the minute you turn it on, you will notice a huge, massive difference in how the QuickTime movie looks on your screen, especially graphics and text. For years I thought my QuickTime movies were doomed to look bad on a computer but good on an external monitor, until a LAFCPUG member pointed out this little feature.

> The appearance on the monitor is not important to me

Garbage In, Garbage Out. If your cut doesn't look good on an external monitor from the timeline, it is guaranteed to look even worse when you compress it and then output to DVD. As Tom and Shane have stated emphatically above, you need to look on an external monitor to judge the quality of your media.


www.derekmok.com

I have never heard -- "BUT...you simply cannot judge the quality of anything using the computer monitor. Never ever ever" so much from any other application.

I use AE, Avid, PPro, XSI, Houdini, Photoshop, Flame, and a host of other applications and I can generally get the image 90% of the way using the computer monitor. Sure if I'm color correcting I'm using a projector or CRT.

But I have never had an image sequence in any [on-line] application where if it was soft on the computer monitor and it looked good on the broadcast monitor. Maybe from an Avid 15 years ago.

What's up with this?

Sorry Tom, I just quoted you because you were the last one to say this. But this is certainly a trend or theme with FCP.

"To:Johan Polham...if you use an NTSC preset sets the pixel size, which means the resolution will only increase by decreasing the file measurements, the size of the file stays the same. This means NTSC at 72 is exactly the same size file as NTSC 300. Increasing the resolution in Photoshop therefore won't help. But thanks for the reminder not to interlace. (I don't think I did, but I'll certainly check next time.)"

-I'm not sure I follow you.
You can change the dpi in the file without changing the pixel size.
They are two different things.
A file that is 1000 X 1000 pixels can have a resolution of 300dpi or 72dpi.
One is clearly sharper than the other when printed.
It only appears to grow in size on a computer monitor DPI is increaed because a computer monitor is set to display at 72dpi. The file does not really get physically bigger or smaller when the DPI is changed.
Putting an image in the timeline that is 300dpi will look better than placing one that has 72dpi.

You bring up an interesting point, however if hedid use a 300dpi image then there should be no way that it appears soft on the computer monitor.

But I think the max an SD CRT can display is 72dpi so I don't think anything above that will effect the quality of the image.

Thanks, good to know, but I was using 300 dpi for all of my tests.

"But I have never had an image sequence in any [on-line] application where if it was soft on the computer monitor and it looked good on the broadcast monitor. Maybe from an Avid 15 years ago"

-As Derek and some other people pointed out, the canvass window in FCP is far from reality. For starters it only displays half the image if it is interlaced.
And if you export as Quicktime movie and then open the file in quicktime it will look fuzzy.
Try exporting a PNG still from the QT movie and open it up in photoshop to have a look.


Having said all this there are better (but slower) ways of working with photos:
After Effects basically always produce perfect results from images.

Johan Polhem
Motion Graphics
www.johanpolhem.com

Chuck...what are your sequence settings? If DV then the quality of the imported image will not look as good as the original...because it has been lowered to the resolution of the sequence. Same with files exported from other applications in higher resolutions. DV ain't the best format to work with.

But, if you make your canvas 100%, you can see the 90% of what it looks like. Doesn't leave you much room to work, but you can.

If you are in AE and working with great images...then yes, it will look good. SAme with other applications. HECK...FCP gives me a GREAT image when I am working with DVCPRO HD. What I see in the CANVAS is darn close to what I see on the monitor. It is DV that you are wrestling with.


www.shanerosseditor.com

Listen to THE EDIT BAY Podcast on iTunes
[itunes.apple.com]

"But I think the max an SD CRT can display is 72dpi so I don't think anything above that will effect the quality of the image."

-Perfectly true in theory but for some reason not in reality.
I really have no clue as to why this is.

"Thanks, good to know, but I was using 300 dpi for all of my tests."

As mentioned above try exporting a still from the QT and open in photoshop.
If it is perfectly clear then there is nothing wrong with your QT file.

When you render in FCP, you render to a specific codec and video format, not a computer format, and not in RGB, unless that's the way you set it. Many other applications, like AE for instance, work in RGB exclusively. If you're working in DV you render to that YUV codec and format. If you work in uncompressed you render to YUV, 8- or 10-bit, using a specific, interlaced codec, designed specifically for display on a video monitor. It is NOT designed to be displayed on a computer screen. The Canvas display in FCP is for preview purposes, and is lores to make playback easier and to allow greater real-time capability. The same happens in the QT player. By default it's set to lores output to enhance frame rate playback. You can change this in the QT player by checking hires playback. There is no similar function in FCP. If you want to see the image as close as it gets to TV display on your computer screen, set the Canvas to view at 100% exactly. You will of course see interlacing, but then that's what you're render out and viewing on a non-interlaced display. At any other setting, the Canvas or Viewer is mushing the interlacing into the screen. Using Digital Cinema desktop is crap also because the image, depending on your computer resolution, is being scaled to fit the screen.

The problem with high print resolution like 300dpi is what happens to the image in the video application. If the image is made to fit the video screen in the graphics application and exactly fits the screen, the dpi will make no difference at all. The problem comes when the image does not conform to the video frame. Then you have what is a high resolution still image, with perhaps a great amount of detail, perhaps more detail than video can support when it has to scale the image to conform to the video screen, perhaps change its aspect ratio to allow for the difference in pixel aspect. Then the codec has to figure out how it's going to mush this great amount of detail and compress it to fit the video specification you're working in. That's when you start to have problems and things that looked clean in Photoshop no longer look clean in the video application, but might look soft, or they might shimmer, or you can even get pronounced moire patterning. It all depends on the content of the specific image. Some images will compress well and conform to video easily, some won't. You have to make the image conform to video and its requirements before you bring it into the editing application. Or take it to a motion graphics applications that's designed specifically to handle this type of material.

> You bring up an interesting point, however if he
> did use a 300dpi image then there should be no way
> that it appears soft on the computer monitor.
>
> But I think the max an SD CRT can display is 72dpi
> so I don't think anything above that will effect
> the quality of the image.

I don't know if this will help or hurt anyone here, but since many of the other wild misconceptions in this thread have been corrected somebody's gotta come in and correct this one before the Martians see it.

DPI (Dots per Inch) has absolutely no meaning in video. Zilch. It is very meaningful in the print world, where halftoned images are set up to occupy a specific amount of flat area typically composed of dead trees. In video it means nothing.

If you're having trouble with this concept, go measure a video inch on your WatchMan and then go measure it on a JumboTron. How many centimeters is it?

Tom I know what you say is correct but equally true for any application that runs on a computer and outputs video. I guess the difference is that other applications have figured out how to correctly convert and display high resolution RGB/YUV in real time.

I'm just venting -- its already been a long week...

Thanks for pointing that out Braker - I was starting to feel a link to a former post so I put it in the FAQs

Please point anyone daring to use the term DPI for video in the wrong context to this article!

[www.lafcpug.org]

*PLEASE ALL OF YOU READ THIS AND DO NOT PROPOGATE THE DPI MISINFORMATION*

______________________

OK CB to answer some of your direct issues.

The reason your output on DVD looks awful is that it is Compressed.

Compressed as 720x480 regardless of source material being larger and better quality.

Not only that but the colorspace is also compressed.

RGB uncompressed PSD files >>> to >>> DVD MPEG-2 YCrCb 4:1:1 (or 4:2:0) files, this loses much of the colour detail but retaining the luminance detail.


You are blaming the applications without understanding the format you are using.

Avid would output this EXACTLY the same and the DVD would look EXACTLY the same at the same settings.

Try this:

Create (in PS), Edit (in FCP) and Output (See below) your files as 1280x720 or 1920x1080

Use Quicktime and the DVCproHD CODEC (to save HDD space and processing requirements) or if you can suffer a little more loss then look at a high bandwidth H.264 encoding.

If you want to make high quality video for output then look into HD-DVD

or

Playback from a laptop or Mac or MacMini via DVI to a HD-Projector or HD LCD/Plasma display.



Regards

Ben



For instant answers to more than one hundred common FCP questions, check out the LAFCPUG FAQ Wiki here : [www.lafcpug.org]

"DPI (Dots per Inch) has absolutely no meaning in video. Zilch. It is very meaningful in the print world, where halftoned images are set up to occupy a specific amount of flat area typically composed of dead trees. In video it means nothing. "

i admit I'm no expert in DPI's etc but it would seem to me that DPI does have meaning in video.
If the files are from Photoshop it will make a difference if the file is 30 DPI or 72DPI.
That makes sense.

What does not make sense is why and 300dpi image looks better in FCP than a 72DPI image.
Does anyone know the answer to this or have I lost my mind?

I have resized an image from 72dpi to 300dpi a few times with noticably better image quality.
(Not only for scaled images but also for still images)

Johan Polhem
Motion Graphics
www.johanpolhem.com

> i admit I'm no expert in DPI's etc but it would
> seem to me that DPI does have meaning in video.
> If the files are from Photoshop it will make a
> difference if the file is 30 DPI or 72DPI.

Well, it doesn't and they don't. Not unless the file also has a different pixel resolution, or is simply a different picture.

> What does not make sense is why and 300dpi image
> looks better in FCP than a 72DPI image.

Are you saying, for example, that a 1600x900 pixel "300dpi" image looks better than a 1600x900-pixel "72dpi" image, from the same source? In other words, the same image, with no change other than to apply the term "300dpi" to it?

> I have resized an image from 72dpi to 300dpi a few
> times with noticably better image quality.
> (Not only for scaled images but also for still
> images)

If you have resized the image, then of course that would make a difference. But changing the DPI count is not resizing. It's an end-product designation for the print world. It is irrelevant in video. Do some reading, some good stuff is resting right above us here.

It seems that people are still confused.

Going from 72dpi to 300dpi in photoshop automatically changes the size of the pixel dimensions and hence, the image.

This is resizing and not a true reflection of the actual detail for the original DPI.

The examples I gave in the FAQ are correct, they concern dpi with 1:1 original pixels and as such make ABSOLUTELY no difference in video or screen graphics, only on the printed output.


Now for your query...

The reason your RESIZED 300dpi version often looks better is because photoshop resizing will add interpolated pixels which add a slight blur to the image and for video "slightly blurred" image actually look more stable. The "detail" that is added is not actually there (in the original).

This is one of the main reasons that in online we apply flicker fixers or a gaussian blur of 0.3 to 0.6 or frameblending to computer graphics or very sharp photos.

There are strict rules for creating good quality graphics for Standard Definition video that is destined for CRT. Many of you won't have ever read about it, but break the rules and/or ignore the techincal specifications and you will end up with a poor or flickery image.

I'll chat with Derek, GFX Joe et al about doing a comprehensive LAFCPUG FAQ / guidelines for creating video graphics and post it soon. In the meantime do a google search as there are many articles from many years on how to create good video graphics for broadcast.

For instance - the recommended minimum line thickness is 3 pixels, any smaller (without blur) and you will undoubtably encounter interlace flicker.

These are the things you learn over many years of training, experiment and working in Online and finishing. Don't expect the average offliner or yourself to be a whizz from day one or that the applications will do it automatically for you (yet).

Sit down and read some tutorials and technical documents on video display technology and then you'll understand why your Red logo on black backgrond looks lovely in PS but ***** on DVD...

Me personally, I love learning new stuff and have 1000s of things I know very little or nothing about, right now I'm eagerly awaiting a "hands on" 4K Red Camera demo and HD transfer and just how much I'll need to invest to cut 2K & 4K in the coming years. As well as all the technical points so I can continue to be of some help on LAFCPUG in the future.

Ben



For instant answers to more than one hundred common FCP questions, check out the LAFCPUG FAQ Wiki here : [www.lafcpug.org]

That's much more useful than my rant, Ben!

As for understanding the relevance of DPI, the real point is that there is no such thing as an "inch" in video. Your graphics get displayed at many different sizes, and you have no control over that.

As soon as you start thinking about video graphics solely in terms of pixels rather than inches or cm. - and ignoring anything you see there with "dpi" on the end, things should become much more clear.

I don't know how PhotoShop comes set up out of the box. If it was set up for inches or cm., I would have changed it right away to measure in pixels. In PhotoShop (CS2, anyway) go to Preferences / Units & Rulers and make sure that "Units" are set to pixels.

In short:

- measure video graphics in pixels
- ignore inches and dpi

First, thank you, thank you! I've been batch downrezzing animation sequences from (big) TIFF files to DV for preview and edit timing decisions only (kinda like workprinting in film) and wondering why they looked awful in QT Player, but not in FCP. Now thanks to setting the playback display in QT Player to' high', they don't. A small comfort, but a comfort nonetheless....

Second, Sythetic Apeture's 'Echo Fire' previews images directly out of PS on an NTSC screen (via firewire.) Saves a lot of to-ing and fro-ing. Cheap, too.

Well I gather from a little bird that PS CS3 will have a specific version for creation of video graphics or that the toolset / display will have a custom setup specifically to handle video. Different somewhat from CS2 and all the disparate settings you have to rig up correctly before your canvas accurately reflects what you will see on FCP et al.

For those wanting to know where 72dpi (for mac screen) and 96dpi (for PC screen) came from, you only have to look at the original Desktop publishing setups and the first WYSIWYG displays that gave you approximately 1 screen inch (when viewing an image at 100% scale) for 1 printed/scanned inch. So we could see the layout as 1:1 before output... (Aaah thems were tha days!... 1bit Video cards at 640x480 and halftone heaven!!!) In fact I used to resize my 22" Mitsubishi's to 1:1 @72dpi so my layouts would be the correct size at 100% scale.

Braker you are quite right - the first thing you should do with your Photoshop is setup for pixels and for the video colorspace you mostly work in.

Although rather than ignore inches & dpi I would say:

"Only use DPI when working on/with printed material or on output or scanned input."

Ben



For instant answers to more than one hundred common FCP questions, check out the LAFCPUG FAQ Wiki here : [www.lafcpug.org]

> "Only use DPI when working on/with printed
> material or on output or scanned input."

Correct As Usual!