I don't understand why you think this is an issue.

Final Cut, Avid Xpress and Premiere Pro- heck even iMovie output from the timeline via firewire by default. I assume Vegas, Pinnacle, Speed Razor, TED and others all do the same at least.

With Final Cut and Avid this output can be at full quality... so you already have this feature. I don't use Premiere Pro, but iMovie *might* drop frames without warning you on older systems (G3's and G4's 867MHz and slower usually), but even that is uncommon.

I only recommend rendering when you have problems with realtime playback, which should not be an issue for any relatively modern system while working with DV or HDV.

The camera can not record better images than DV or HDV anyway, so SDI is overkill for these applications. The camera will just throw away extra data.

If you get a deck with SDI input as an option it usually is a format better than DV or HDV anyway, which CAN use at least some of the extra data. The few DV decks that take SDI in just throw away all the extra data, just like a camera would. Same thing for DVCAM and DVCPRO. Maybe a few "golden eyed" viewers could detect the difference in your masters, but they certainly won't notice once you run out to DVD, streaming or broadcast or wherever your footage ends up.

To sum up, SDI inputs for DV and HDV systems are worthless from a quality standpoint.

The only facilities that need this are ones that have standardized on SDI for other workflow reasons, and it is there for convenience rather than technical reasons. (SDI is ready, so you don't have to go find a FW cable, etc. etc.)

SDI output from the camera is MUCH more important. If you need higher than HDV/DV quality then get a deck which uses a better format. DVCPRO 50, D-9 (both DV25) and DVCPRO HD come to mind.

Is the output with firwire for HDV from the timeline live without any rendering? I was under the impression that HDV worked like DV where any effects needed to be rendered first before they could be previewed through firewire.

I thought this was one of the big points for people to edit HDV with a Decklink card so they could get a realtime preview (not a record to tape) to an external monitor.

I didn't mean there would be any quality boost. I am only looking for a way to preview my edits without rendering. While editing DV on an uncompressed system may not gain any extra quality it does help in terms of realtime performance and feedback which is why a lot of people do it.

Am I wrong in that unrendered effects will not play through firewire from the timeline?

1) Single 35mm CMOS chip.
2) direct hook up to Canon 35mm EF lens- no mag factor.
3) Super low light handling without apparent noise even at +18db.
4) New viewfinder with double density LCD and double brightness.
5) Smooth iris wheel. No visible stepping.
6) Superior A to D audio circuitry w/ 4 channels of up to 24b/96khz.
7) Built in individual phantom on all channels, XLR connectors great mic pres
providing tons of clean gain and stepped attenuation pads for
both -10 and +4 db.
8) True progressive recording in NTSC, PAL, and HD formats 720P, 1080i and
1080P@ from 4-60 FPS.
9) 10 hour battery (5 hours of real world recording no problem)
10) Up to 10 seconds of pre recording to RAM.
11) State of the art memory pack capable of 1 TB storage and instant
select, playback and delete.
12) Mark "good take" button next to record button.
13) All flavors of digital video input to memory pack.
14) Digital still with onion skin option for re-setting scenes.
15) High power (10 watts) headphone amp.
We'll start with these :)

Yes, you CAN play video back over Firewire from the timeline.
 

Most, if not all, modern NLE software will output DV and HDV live over firewire right over the timeline. Final Cut Pro will also output DVCPRO 50 and DVCPRO HD over firewire, but these require the firewire device (camera or deck) to "understand" those formats. (So no DVCPRO HD to the XL-1s...)

My worklow in my home studio using Final Cut, or iMovie is this:

Plug in Firewire video device (usually my JVC SR-VS10U, but I did have a HD100 for one small job)
Start the edit application
Capture video.
Edit with realtime timeline display on my TV.

Its pretty much always that flawless since Premiere 6 was new. Even on a PC. (Yes that's a troll.<grin>)

Analog or SDI outputs from the NLE are meant for two things.

1) Output of arbitrary timeline content as pure video. With firewire outputs, you must be outputting a specific format to a device that understands that format. This is why some video clips need to be rendered before they can play. More on this later.

2) Output to devices that don't have onboard video processing capabilities. Direct output to a monitor is an example, as opposed to a camera which can interpret DV input, which is just data, as video and route it to a video out.

You might notice these are minor variants on each other.

Final Cut Pro and Avid Xpress can output HDV direct from the timeline over firewire. I don't know what Premiere Pro can do, and I personally don't care, but I'd be very surprised if it could not. iMovie can do it, so I assume it is par for the course with any NLE software now.

As I said before Final Cut can also output DVCPRO 50 and DVCPRO HD over firewire entirely in software. I think that Avid Xpress requires a Mojo box for those formats. <shrug>

Using FCP or Premiere when you have DV on the timeline you can see it via firewire outputs without any rendering. Depending entirely on the speed of your system some effects may play back in realtime from the timeline without rendering. Straight cuts and crossfades usually playback in RT on modern systems without rendering. Other effects depend on your software/hardware.

Some hardware accelerates this. For example Matrox RT series cards and Avid Mojo do some video processing in hardware, and as a result some effects playback in RT without drawing CPU resources. The latest release of Avid software on the Mac uses Apple's Core Video technology to make the video card render stuff in realtime which it does REALLY well. Pinnacle Liquid does this too, using Pinnacle's proprietary video card interface though. Apple's motion uses Core Video and is ridiculously sweet, I can't wait for other FCP Studio apps to get on board.

Most software uses hardware rendering to allow more realtime features, by using the CPU to render effects the hardware can't.

Now there is a big gotcha with all this. Your video source clips must be in the format you are set up to play out the timeline. So, if I am editing DV, but I have DVCPRO clips on the timeline I will have to render. If my timeline is uncompressed HD and I have DVCPRO 50 on it, I'll have to render. etc. Every clip should be in the format you have selected for your timeline.

Some high end stuff will let you mix formats on the timeline without any rendering. Avid Symphony is an example. So is Avid DNA. I think Sony Xpri does it too. So does some Pinnacle Liquid stuff. (Avid bought them for a reason!) I haven't been keeping up with all the different vendors so I defer to those who have.

One nice feature of some high end Avid systems and the Video Toaster's TED is that they render in the background. So, while you are doing your thing, choosing edit locations etc., TED and DS|Nitris render anything that won't play in realtime with any spare CPU power available.

I've been lobbying for that feature in FCP Studio. DVD Studio Pro has it, but that is it. If implemented in a fashion compatible with Apple's Xgrid clustering a relatively cheap Apple cluster might smoke some high end and expensive Avid hardware.

Well, I am getting off topic. Sorry.

Uncompressed video source material is much easier from a processing perspective to add effects to. This is because compressed video, like DV and HDV, need to be decompressed, have the effect done, then recompressed. (Just the way the math works.) Uncompressed can be directly manipulated, freeing up quite a few CPU resources. So uncompressed is easier to work in without rendering.

There is a HUGE caveat though!

Uncompressed suffers from the "minor" nit that computers have a hard time keeping up with it at all from a disk bandwidth perspective.

For every stream of uncompressed SD video your system can handle, you could instead be playing back ~10 streams of DV.

For every uncompressed HD stream your disk subsystem can handle you could instead be using ~14 streams of DVCPRO HD, or 60 streams of HDV.

(CPU limits bound this at about 10 streams I think. DVCPRO HD and HDV have the same realtime playback performance. H.264 HD material can bring a fast machine to its knees, so only expect to play maybe 3-5 streams simultaneously, despite low disk throughputs.)

I hope that clears some things up.

Interesting list, I can't resist commenting...
 

I have a hard time understanding why Canon didn't attempt this to begin with. I brought it up years ago on USENET, and to every Canon rep I could find.

Even with a magnification factor it makes great sense for Canon. The current 7.2 magnification is crazy. Trying for 1.0 magnification would be expensive, but it is reasonable for them to try to aim for a 1.5-2.0 mag factor.

They already make a lot of lenses, so economies of scale could help them reduce costs.

If there are no more manufacturing economies, then prices would go UP increasing per unit profits.

They would still have to make video specific lenses, because video lenses need features not typically available in still lenses, like ND filters.

35mm still lenses may not be perfect, but they can handle HD video very very nicely. Full resolution 1080p HD is only a 2Mpixel application and their current lenses resolve much finer.

I guess Canon thinks it can't risk upsetting its pro video customers with that type of video camera.

I think that is wrongheaded. per unit profitability is very high for Pro lenses, especially HD ones. Still that is nothing compared to what they make from still photographers. Some manufacturers would continue to use Canon lenses.

If Canon won't try this, then perhaps Nikon will enter the video fray with a camera based on this idea. They make great stuff too...

There are engineering problems with this. Big ones. If they COULD do it, they certainly would have tried at least for their professional markets.

Even in an ideal world the lenses limit this greatly.
Cost, cost and cost.

Still, given the difficulty of accurately focusing HD video I think we'll see some offerings in this area. Still, don't expect miracles. The viewfinders on Sony CineAlta cameras are hard to focus HD with.
This is really a request for a mechanical IRIS just like on pro cameras/lenses.

I have asked for this too, but cost is a huge issue. Digital iris is "almost free" to implement.

I suggest simply adding more steps to the iris controls. Then use a larger iris wheel. Detente's at each full stop of aperture would be a must, as would fixed motion. I suggest a maximum of 120degrees rotational freedom, with maximum wide at one end and closed at the other. The camera would have to move the ring during shutter priority or automatic operation, so it would be in the correct place if the user changes modes.

Reliability of such a mechanical device is an issue.

Still, I think it is a GREAT idea.
I think they haven't done this mostly to stay in compliance with DV and HDV specs.

If a future Canon HDV camera had this feature and you used it, you would not be able to interchange tapes with other HDV cameras/decks. Also your NLE would choke on it.

You can find this feature on pro cameras.

In the meantime it may be easier to just use 16bit 48KHz or record external audio using DAT or some other device.

We could hope for a new standard for video...
Good idea, but engineering trade offs. Doesn't the H1 provide phantom of two XLR's ?
The "F" mode on the H1 seems to come VERY close to this according to Chris. I would like to see a whole crop of cameras that support every HDV standard out there though, that seems to a good idea for the entire industry and the standard.
There are battery systems that provide this without any problem. Cost cost cost. Anton Bauer is a good example. If you check out Pro cameras that use these battery systems, they don't typically come with batteries, and the batteries are EXPENSIVE.
GREAT IDEA. It will increase the cost, but be worth EVERY PENNY.
That's fantasy land for quite a few more years. How's about 0.1 Tbyte ? 100GB 2.5" laptop drives will be available really soon. In swappable packs they give about two hours of DV100 (DVCPRO HD) recording. They aren't outrageously priced like the Flash RAM in P2 cards. You could reasonable expect to own ten of these. Especially if there is a standard for their enclosures.
I don't know if having it as a button so close to record is a good idea. I can just see myself and a million others marking a take as good instead of hitting record. I know that despite all care and precaution I'll do that at least twice.

Take review/marking is a good feature. It is easy enough to implement with DTE HD systems, but it should be kept away from the regular camera controls. Also, the slight inconvenience of doing so gives an opportunity for a better interface design that allows more flexibility.

Let me take it a HUGE and unprecedented step further:

Include integration with a palmtop device and Wi-Fi. Use the palmtop not just for marking takes good/bad, but for logging in the field.

It would be BRILLIANT in the field. As a producer, director or AD I could take notes on the Palm between takes. The camera could share take data, like a thumbnail and time/timecode. With a permission system I could authorize one user to write these notes onto the camera media, and others to have their notes stored on the Palm or on a central database.

That's HUGE. Multiple users being able to comment on takes while on set in a permanent way that can be used to edit with later!

The same system, while intended for use in the field with palmtops could also be used with laptops and desktops which may already be on the production location!

With some smart ID tagging of media, you could have an EDL ready before you even plug the media in to the edit system. With a DTE device, editing might take seconds. Absolutely KILLER for ENG applications.

This is one of those ideas that's worth a few million dollars. If camera/NLE/computer makers catch on to this and deliver we'll wonder how we ever worked without it.

Hello ? Anybody out there who makes these decisions ? Write me now and let's get the ball rolling!

To the memory pack ? Why not.

Getting a camera that can do that is a while off. Remember these things have very low power on board processors. We have talked about applications that will require adding one or more general purpose processors into a power sensitive device.

Still, if a standard emerges soon enough for DTE media there is no reason why you couldn't take your media from a DV camera to a DVCPRO HD camera an then to a DS|Nitris.

What do you mean ?
Too much power for headphones I think. We really need better noise isolation and active noise canceling combined, not more power. I think it'd be the same effect, just a different way to get there. One that's easier on my ears.

Quote:
Originally Posted by Jacques Mersereau
14) Digital still with onion skin option for re-setting scenes.

What do you mean ?
___________________

Onion skinning is where the still image would be shown on the
camera's ouput and given translucence of say 50%. Under that is
the camera's real time image. This means if you need to reset a scene's
elements like props and/or people placement you can with a high
degree of accuracy. This can also be helpful in achieving special effects.

BTW, most of those HD2 improvements I suggest are pipe dreams.
For now, I hope I will be happy with the HD1, and being a wildlife
documentary maker, I LOVE the 7.2 mag factor of the HD1. An HD2
without out that would be another great tool in my kit.

It would be so easy to add onion skinning to this camera... Canon already offers a "title mix" function on their consumer one-chip camcorders. The two features are closely related.

why no DVCPRO HD?
 

I don't understand why canon aren't offering DVCPRO HD as an output from firewire. surely the DSP can handle it if it's already doing HDV compression...

As people have already pointed out without a standard such as this you're a bit stuck capturing HD in the field. why spend $9000 on a camera that's limited to HDV outside the studio? are they going to wait another year and put DVCPRO HD into a XL H2?

It just seems like the ability to strap a little box with a HDD in it around your waist or to the camera and record a HD format less compromised than HDV but within the limits of a single disk via firewire would make the H1 so much more attractive to a wider audience... so why aren't they doing it?

Hi Joe,

"I don't understand why canon aren't offering DVCPRO HD as an output from firewire."

There's one tiny little snag that prevents Canon from offering DVCPRO HD. They don't own that format. DVCPRO HD belongs exclusively to Panasonic. Panasonic would have to choose to make DVCPRO HD available to Canon, and Canon would have to pony up for the licensing fee. Neither company may be willing to do that.

On the other hand with HDV, Canon is part of that consortium so they have the right to offer it.

As far as recording DVCPRO HD to a portable hard disk recorder, the recorder would need the ability to handle the bit rate of that format (the FireStore from Focus Enhancements will have this capability) and you'll need a portable Panasonic DVCPRO HD encoder between the XL H1 and the recorder (and none yet exists that I'm aware of).

DVCPRO HD is a mechanical system, tape standard etc. etc. that happens to use the DV100 codec for HD resolution material. Its just the DV codec at 100 Mbps at HD resolution. (Well, its four DV codecs ganged up. I think it was developed by Panasonic and JVC, but they might have been working independently. <shrug>)

What I am getting at, is that while Panasonic definitely owns DVCPRO HD, they may not own the codec. If they do, they really just own a specific implementation of the idea.

While Canon might never be allowed to include a DVCPRO HD mechanism or even the codec, they could certainly produce a 100Mbps HD codec with 4:2:2 sampling for 1080i and 720p formats. Like JVC and Panasonic did they could base it on the IP from the DV consortium. (HDV is not recyclable in this case.) Alternatively they might be able to design a 4:1:1 codec using intraframe compression at 50Mbps.

So how hard would that be ? I am not apt to try this, but its doable.

I would immediately abandon the notion of using a tape mechanism though. They are expensive to design with reliability. DVCPRO HD moves tape pretty fast, so does HDCAM.

I also would not try Optical disc media. A blue laser optical disk the size of a DVD could hold enough data for one hour of 100Mbps recording, so its satisfactory from that standpoint. It is also direct-to-edit (DTE) capable. You are still left with power and mechanical issues aplenty. Try recording a DVD-R at 16x and you'll HEAR the issues.

I would choose hard disks.

As Chris said you can definitely write this out to DTE hard drives. We are about to get a crop of 2.5" SATA 2 100GB HD's with 16MB cache at 7200 RPM. Writing a 100Mbps stream would not tax these drives at all. Laptop drives are very well suited to this application mechanically. Also, two hours of recording is attractive.

In many applications even using 3.5" drives would be acceptable, so large capcity drives running up to ten hours are possible. (Basically anywhere you are taking a huge A-B battery pack the size weight of 3.5 vs 3.5 drives are negligible.)

While they are mathematically simpler DVCPRO HD, DVCAM and similiar codecs are more difficult to code than HDV computationally. What I mean to say is that the equations are simpler, but they take more computer cycles per frame to actually calculate.

This is because they have to encode all image detail using only a single frame. That means all image detail must be encoded/decoded with every frame in real time.

With HDV and any other interframe codec part of the math for each frame has already been done. Also much more of each frames data is discarded. That results in a lower calculation requirement per frame, and a lower data rate.

The point of that is that you need a fairly modern and fast DSP to handle coding for such a proposed system. Both Freescale and Intel make a bunch of embedded CPU's that can handle the task today. Personally I'd choose the PowerPC based chips because that architecture lends itself to multiprocessing and parallelization if it becomes needed.

I think that 100Mbps is a "magic" number. I do not know that you can do intraframe compression for HD video at less than this data rate. I fairly sure you can't do intraframe at 4:2:2 at less than 100Mbps.

Take that last bit with a good size grain of salt: I haven't studied codes and cryptography in about a decade. (Mathematically the same topic.)

Yeah, that would be a great function and VERY easy to implement. They really should do this immediately. Maybe even for the H1 if its hardware isn't set in stone yet.

I am not a big fan of attempting composites in HDV, but it certainly can be done. Still for a camera that is intended for high quality studio applications (that's what SDI out really means for now) this is a killer feature.

This suggests another feature to me though: A brief framestore in uncompressed HD. 1-2 seconds is all I'd suggest. You need 256-512MB RAM for it. This could be used in a large number of ways. For one you could take a still, store it in the framestore and use it for onion-skinning.

LOL.

I agree, they are pipe dreams. Some of them are more so than others. Some are just out of reach, some can be done right now, others either are misfeatures or already available.

The thing is this: I believe that people from the manufacturers R&D and engineering departments lurk in pretty much every public forum. It is in their interests. They pay attention to off the wall threads like this because they need feature ideas.

I believe (well its more like blind faith) that if we post reasonable and well thought out ideas we may see them in future products.

So, in re-reading what I wrote in response to you I have to say that I wasn't pooh poohing your ideas. A lot of them are very good and needed in cameras at the price points we work with. I am just trying to add in a bit of armchair engineering. Hopefully enough so they get off their duffs and say "Hey we CAN do that pretty cheaply!"

As to the 7.2 mag factor being a feature of XL camera's. Well I have to admit it can be a lot of fun and has some utility. Doubly so for what you do.

Most of the time however I end up needing a wider wide angle. I think 35mm still cameras in kits come with 28-104 lenses these days. That field of view should be a standard for video cameras.

Sticking with XL-lenses I think the standard should be a 20x 3.8-76mm. We should get a 4x 2.5-10mm wide angle lens, and a 10x 20-200mm.

A move to 1/2" CCD's would be very welcome instead.

i don't get people's obsession with recording to the hard drive. it's just 2 inmobile. at 1GB a minute of 1080p (roughly) 1TB will yield you 16 hours. so you can theoretically take 2 of the 500GB SATA2 desktop hard drives and make them raid0. the problem is weight and portability. so while it is technically possible, the point of XL series is high functionality+portability. and don't even think about solid state yet. i give it another decade or two before it is cheap enough for practical use.

realistically, the tape transport is still the thing to beat... but i wouldn't follow an existing standard. i'd design a tape system from the groundup that ain't as small as miniDV but not too big (like VHS) but enough to hold 25mpbs 1080p (variable fps) for 2 hours or more. why record @100mpbs when HD-DVD or BluRay will only do 25mpbs????? what's the benefit of 100mpbs besides wasting bandwidth?

thanks Chris and Alexander for the format info.

It makes perfect sense now why the H1 doesn't do DVCPRO HD.

But as Alexander says they could still provide something equivilent. It just seems like a really incongruous mix of features: HDV, SDI-HD, nothing in between... Sticking with tape I suppose unless they developed a new standard and possibly tape format they are stuck with HDV, fine. But why not as do as others have suggested, double or triple the data-rate and put that out through firewire for a FireStore type solution? So the CPU might be working a little harder and use more battery, not the end of the world and certainly worth it for the feature, surely?

Yi Fong, I'm by no stretch of the imagination an expert but I believe the benefit of shooting at a higher datarate than distribution it so you have the option of processing the footage more before it starts to fall apart. If you're doing things like compositing then you'll have to compress twice to get it to DVD (or whatever). Once on capture, once again for distribution after processing. I don't know if going through 2 19mbps MPEG2 compressions is roughly equivilent to once at 9.5mbps but it's certainly less than once at 19mbps, which from the m2t footage from the JVC HD100 I've seen here already shows noticable artifacts.

I guess I'm just a little frustrated that the H1 is so close to what would be realistically ideal for me but just lacking a format that can go straight onto a single HDD and that I don't have to worry about when I postprocess. damn. The HVX200 will have the FireStore, if only it had interchangable lenses... maybe the H1 firmware can be mod'd to give a higher datarate... but of course then there's the issue of making your NLE understand it and a FireStore record it. It's looking to me now like the HVX200 is probably my best bet.

then there's a simple solution. redesign the tape transport to handle uncompressed HD image. after all, VHS was made to handle uncompressed audio before, there shouldn't be a reason why this kind of engineering couldn't be done. magnetic tape is still much more cost effective than any other medium out there.

DVCPRO HD=100Mbps=12.5Mbytes/sec=750MB Minute.

That's ~43 GBytes per hour.

A 100GB laptop (2.5") drive will store 2 hours.

Right now such a drive is expensive, but in 2-5 years it will be cheap.

They are very mobile, being far smaller than VHS. (Heck even 3.5" desktop drives are smaller than VHS, and about the same size as Beta for that matter. Got all three on my desk right now.)

I think that those facts obviate the remainder of your argument regarding size.
Why not ? Panasonic has with their P2 system. Apple has also delivered 4 GB devices by the tens of thousands. OK, so that's only about 5 1/2 minutes, but it is out there. I think we'll see 16GB flash drives by summer 2006, and 32GB devices by the winter. We are likely to see 64GB devices in late 2007, and 128GB in early 2008.

I am not being particularly optimistic either, since Samsung released 16 Gigabit NAND Flash chips about a week ago. You need eight to make a 16Gbyte device, but that is standard. Samsung expects to beging mass production in the second half of 2006.

http://news.yahoo.com/s/afp/20050912...g_050912043923

That depends entirely on your definition of cheap enough. Given how much I spend on tape I am looking for options.

Optical is CHEAP, but I need a fast way to get video onto optical disks.

Oh, I've said it before and I'll say it again: Optical has all sorts of mechanical issues in mobile applications. Making it reliable for camera use is NOT easy, or cheap.
Congratulations, you have just specified HDV once again for the first time.

In case you missed my sick sarchasm (sic) HDV not only meets but exceeds your specifications.

Also, tape transports are incredibly complex systems. I suggest getting an VHS deck and a service manual for it to see what I mean. Break the deck a few times, then fix it to learn. Remember- that's a consumer technology that's over a quarter of a century old.

When you deal with tape transports on camera's they become even more difficult. Just take a look at the tape transport in a typical DV camera.

Not easy.

That's why I keep suggesting hard drives. There are a lot more hard drives made every year than there are video cameras and tape decks of all sorts combined. Every year the number of drives on the market grows faster than any other electronics, except processors.

The trend for the last five years is that a greater and greater percentage of all computers (and thus hard drives) sold are laptops. Lapop hard drives have special features of interest to people with mobile applications, like say video recording.

Apple has also been doing a rather spry business with mobile hard drives on this iPod thing of theirs, you may have heard. It uses 1.5" hard drives in capacities up to 60GB... and they are fast enough for video recording.
Uh... because greater bandwith recordings have higher fidelity.

You want to create your materials in the highest available media fidelity. In other words you want to work with the best images, so when you send your work out to end users it still looks good.

HDV is a 25MBps signal. IIRC it is a 4:2:0 sampling, and very highly compressed. That is lossy interframe compression with long GOP MPEG-2.

DVCPRO is a vast improvement in quality. 4:2:2 sampling at 100Mbps. The compression is intraframe.

Advantages of DVCPRO HD over HDV ?

DVCPRO HD can be more easily and more "cleanly" composited.

DVCPRO HD is easier for any video system to jog/shuttle in forward or reverse.

DVCPRO HD is less suscepible to media errors. (Every frame has all the data needed to display it. With HDV you need several frames to create any frame. If any of the needed frames is damaged you lose all the frames depending on it.)

DVCPRO is very clearly better looking.

Here is an article which sort of covers the basics,
http://www.creativecow.net/cgi-bin/p...mer/index.html