USA   F-111 Aardvark


Republished by with the express permission of Carlo Kopp.  More articles here.

F-111 Upgrade Options

Part 2 Performance, Signatures and Radar Warning Receiver

by Carlo Kopp
Published in Australian Aviation, 1998
Copyright   (c) 1998 Carlo Kopp

Part 1 of this feature explored existing upgrade programs on the F-111, and identified capability and survivability issues in relation to the evolving regional environment and technology base. In this second part we begin our exploration of potential follow on upgrades to extend the F-111 through to 2020.

Given the changing broader regional environment, developing technology, and identifiable limitations in the existing and planned capability package for the F-111, we can in turn identify a series of specific upgrade measures to extend the aircraft tactically for the latter two decades of its operational life cycle.

Survivability Issues

It must be noted that survivability and offensive capability can often not be easily separated, in the sense that particular weapons and their delivery profiles may enhance both concurrently. Better basic defensive measures applied to the aircraft will enhance its offensive capability by allowing it to be used in conditions which would not be otherwise viable, and to target munitions more effectively. Increased lethality reduces the number of repeat missions which might need to be flown to kill a given target, and thereby improves force survivability in a sustained campaign. So the often peddled argument that money is better spent on purely defensive measures inherently misrepresents the issues.

To address the survivability/lethality issues, we will need to focus on the four S', Speed, Signatures, Standoff weapons and Sensors.

Speed and Sustained Speed

Speed is vital to the F-111 both in the penetration of surface defences and the evasion of fighters. When penetrating at any altitude, speed denies the opponent response time, and every second which is taken away from a SAM system operator is another second they could use to set up and conduct an engagement. Fighters have finite fuel capacity, and the demands of high speed intercepts, be they off a runway or from a standing patrol, limit the fighter's time to engage the F-111. In many situations, the F-111 will simply outrun the fighter since it can sustain supersonic speeds at any altitude, including 200 ft AGL.

Sustained high speed flight in the F-111 bites into fuel reserves, and the central issue is therefore how to minimise fuel burn at high speeds, or how to ensure that additional fuel is available.

Three measures exist which can be used to improve high speed endurance. The first is the obvious answer, which is inflight refuelling. This means operational tankers in sufficient numbers to support the SRG. Ideally these would have booms, since the boom provides a much higher fuel transfer rate than a fighter sized probe/drogue package. The alternative would be to fit a probe to the F-111, and replumb the aircraft accordingly, accepting the fuel flow rate, drag and signature penalties of a fixed probe arrangement. It is unlikely that space could be found for a retractable probe, and flight testing would be more expensive.

While a robust operational tanker force would be highly desirable, it is not the only means of improving aircraft penetration speeds. The other measure is to minimise the aircraft's drag. The drag contribution of the F-111's four external swivel pylons and stores is considerable, and drag once stores are released is often hardly better, due to the aerodynamics of the pylons, under some conditions. Empty pylon drag is typically of similar magnitude to the drag of the store.

Indeed the FB-111A/F-111G has jettisonable pylons for this reason, the intent being to shed the pylons retaining only a short stub once weapons have been delivered. The option now of course exists to strip the jettisonable pylons from USAF F-111G stocks, and modify the C-model to accommodate them, accepting that these will need wiring changes to accommodate the AUP system's defacto Mil-Std-1760 interfaces, and also accepting that additional stocks would need to be built up if this is to be practiced operationally.

Internal Weapon Carriage

The resolution to the external stations drag problem is simple - use the internal weapon bay to carry munitions. At this time the internal bay is used for the Pave Tack, and the six weapon station bus decoder boxes (each in a slimline chassis designed to fit inside a Hornet pylon) . Pave Tack is simply not required for the delivery of GPS guided bombs, glidebombs or missiles. So should the weapon bay be activated, and GPS guided bombs, glidebombs or missiles be carried internally, then the aircraft can penetrate clean and exit the target area clean. Moreover, external tanks can be carried to increase the unrefuelled combat radius, and these are a relatively low drag store, compared to Multiple Ejector Racks or many munition types.

The Harpoon would be a very attractive candidate for internal carriage since the missile is relatively draggy, and imposes speed restrictions on the F-111, as it was designed to be carried by the subsonic USN A-6E and P-3C aircraft. An air launch Harpoon fitted with folding wings, as used on the tube launched version, fits easily. The standard air launch version would appear to fit with very minimal vertical wing clearance. Both versions would require an adaptor shoe to offset the MAU-12 position suitably.

The F-111 was built to the USAF SOR-183 requirement, and its primary design specification was to penetrate low and fast carrying a pair of internal nuclear weapons, either the B43, B57, B61 or AGM-69 SRAM. The Mk.84 2,000 lb and M118 3,000 lb bombs were cleared for internal bay drops by the USAF, but never used operationally from the internal bay. The weapon bay hardpoints have therefore been built for such loads. While clearance/separation testing will be required for weapons such as the GBU-31 JDAM, their similarity in aerodynamics, size and weight to the standard Mk.84 indicates this will not be an issue.

The F-111C aircraft bays have had the MAU-12 ejectors removed, the F-111Gs retain them. Activating the weapon bay on the F-111C and enabling stores control access to the stations is a relatively simple engineering task. The stores decoders must be relocated from the side of the bay into the upper rear bay, clear of stores and the Pave Tack cradle, and a cabling harness is required to couple them to the cable entry point into the weapon bay. Basically this is a wiring and sheet-metal chore. If the aircraft is to have six rather than four smart weapon stations, two more decoders need to be added and the stores control OFP (software) tweaked accordingly. But this modification can be done even more cheaply, if we choose have only four smart stations, which is not an unreasonable limitation from an operational viewpoint.

Two simple solutions are possible. The first is the simplest, which is to bring the cables from the internal and inboard swivel pylon stations to a bracket mounted bulkhead connector, and then select the station by plugging the decoder into either, using a short cable harness. The unused connector terminating the cable harness would be sealed with a simple screw on cap. The alternative is a switching arrangement mounted in the weapon bay, used to reroute the signals from the decoder boxes, to select either the inboard swivel pylon stations, or the internal weapon bay stations.

No changes would be required to the software, and the total installation cost will be confined to sheet-metal brackets, cabling harnesses and at most a some milspec switching hardware, or another simple sheetmetal bracket. Operationally, the Navigator needs to only remember that his inboard stations are in the weapon bay, and the ground crew need to select the stations accordingly during weapon loading. Neither are out of the bounds of existing operational practices.

The only cost issue is that clearance testing will need to be carried out to determine the limitations of the installation. Since the USAF has already done much of this during early testing of the F-111, for standard stores such as Mk.84 series bombs, the process will be much simplified for aerodynamically similar stores.

Whether the "expensive" six station approach is adopted, or the "cheap" selectable four station approach is used, the engineering overheads of this exercise are minimal and arguably well worth the effort in terms of performance to be extracted from the airframe. It should be noted that this modification does not in any way preclude the existing Pave Tack installation, the choice is simply whether to fit Pave Tack and carry external laser guided bombs, or remove Pave Tack and carry internal GPS guided bombs. The swap-in/out operation for the Pave Tack cradle and pod takes about a day to perform, and involves mounting the cradle and hooking it up to aircraft systems (hoses, cables).

A useful and cheap enhancement to this modification is to provide automatic bay door opening under software control, similar to the AUTO mode in the F-111D/G. This is another low cost modification, involving primarily wiring and a minor addition to the mission computer software. Scheduling the doors to open automatically 2-3 seconds before the drop, and close immediately after the drop, minimises the time during which the open bay compromises the aircraft's signature, while also reducing crew workload during the critical delivery phase of the mission.


The third measure which can be applied to increase the aircraft's penetration speed and endurance is the fitting of a much better engine than the sixties TF30 series. The existing and funded engine upgrade from the TF30-P-3/P-107 to the TF30-P-108/109RA is in many respects a partial measure, aimed wholly at achieving spares commonality across the fleet, gaining a few percent of additional thrust as a byproduct. It is however a very cheap upgrade to perform, using boneyard hardware, and thus could be approved without the major internal political dramas associated with getting funding these days.

The case study for an upgrade from a TF30 to a contemporary engine is the USN's F-14A+/D upgrade to the GE F110 engine (see table), replacing an engine almost identical in performance to the TF30-P-108/109RA being fitted by the RAAF. This upgrade involved primarily the insertion of a 1.27 metre annular plug behind the engine core to match the fuselage tunnel length between the shorter F110 and the longer TF30, the fitting of an inlet adaptor to match the slightly smaller fan to the inlet tunnel, secondary structure changes at most, the rescheduling of the variable inlet ramp, and the necessary flight testing. The F110 series is almost identical in weight to the TF30, and the commonly used F110-GE-129 IPE variant delivers 17 klb static dry thrust, and 29 klb static reheated thrust.

The F-14A to F-14A+ (redesignated F-14B) upgrade produced remarkable results in aircraft performance, with a 43% increase in reheated thrust, a 37% increase in dry thrust, 30% lower fuel burn in reheat, 62% greater intercept radius, and about a 30% improvement in endurance on station and combat radius, all with a much more reliable engine with unrestricted handling. Variants of the F110 are now widely used on the F-16C, are the standard powerplant on the F-14B/D, and have been tested on the F-15E.

What would such a powerplant do for the F-111 ? The first and most important improvement would be in the ability to sustain fast transonic or supersonic dry cruise for much longer when penetrating at any altitude, within Turbine Inlet Temperature (TIT) limits for the engine. A pair of late model F110-GE-129 EFE engines would provide the F-111 with about 50-80% more thrust at maximum continuous power rating, Mach 0.75-0.8, in comparison with the currently being fitted TF30-P-108/9s.

In most regimes the engine Thrust Specific Fuel Consumption (TSFC/SFC) of the F110 is superior to the SFC of the TF30. The F-111's best defence against fighters has always been to outrun them, and with F110s fitted this tactic becomes much easier to apply, since the aircraft can sustain high thrust ratings much longer on any given fuel load.

An important issue in evading the latest generation fighters is that of avoiding detection by Infra-red Search and Track, thermal imagers, or Night Vision Goggles. Use of reheat is highly counterproductive, in this respect, since the speed gained is at the cost of detectability. Even basic NVGs will allow visual detection of an aircraft in reheat at many miles of range. If the F-111 fitted with F110s can achieve the same dash performance on dry thrust, as the older TF30 delivers on low reheat, then it can run at high speed for much longer without the signature penalty of an afterburner plume. So the benefit of sustained speed is achieved, yet the aircraft is not more detectable than it currently is running on dry thrust.

The second aspect of higher installed thrust which is of some usefulness tactically is that the aircraft can sustain much higher G in turning manoeuvres, without the need to light afterburners. Higher thrust to weight ratio improves sustained turn rates, a problem area with the heavily loaded F-111 even at optimal wing sweep settings. While this will never turn the aircraft into a dogfighter, it will improve the aircraft's basic manoeuvrability and this is always useful, particularly in evading SAM shots.

Fitting an F110 variant would produce other benefits:

  • A combat radius improvement will result, since the TSFC for the F110 series is better than that for the TF30. If the F-14 experience is applicable, anything up to 30% could be achieved - suggesting a combat radius up to 1200-1400 NMI, subject to loadout and profile. This is an important strategic dividend, especially in a period where the combat radius of potential broader regional opponents has almost tripled, with the deployment of the Su-27/30. It will significantly reduce the demand for tanker support, in turn reducing tanker force numbers and operating costs.

  • The F110 is designed for operation with poor inlet airflow, and with a Full Authority Digital Engine Controller (FADEC) the engine has a reputation for carefree handling under all flight conditions. Therefore, the throttle restrictions of the TF30 on the F-111C would vanish.

  • The F110 is a newer and much more durable engine than the TF30, and since it is widely used on the F-16C, it will be much easier to support in the longer term than any TF30 variant. The total manhours expended on maintenance will be reduced, and the spares stock requirements will drop. The unscheduled shop visit maintenance rate for later models of the F110 engine is about 1.8 per 1000 hrs. Moreover, since the engine is common to the USAF F-16C, obtaining spares at short notice on remote deployments should also be much easier.

  • The F110 would provide much improved hot/high takeoff performance at higher takeoff weights, thereby increasing flexibility in the choice of basing (although arrestor cables can be installed for shorter runways to provide for safe aborts). The extra thrust would prove particularly useful when carrying a pair of draggy 3,000 lb AGM-142 SOWs and their Datalink pod.

  • The F110 would permanently solve the ongoing issues with the provision of adequate bleed air capacity to support current, and particularly growth onboard aircraft systems. An example is the Pave Tack, which consumes a large proportion of available bleed air.

While to date bleed air has not been a critical operational constraint for the RAAF, the USAF had endemic difficulties in this area. The original P-3 on the EF-111A could not deliver adequate bleed air to support the Environmental Control System (ECS), as well as the cockpit pressurisation, weapons bay cooling, overwing fairing seals, hydraulic accumulator pressurisation and other aircraft systems, under all operating conditions (since bleed air is drawn at the expense of engine performance). This produced a requirement to shut off the bleed air for take offs, to achieve acceptable single engine takeoff climb rates, and turn it on at 250 KT after takeoff. While the fitting of the P-109 to the EF-111 alleviated this problem to some degree, it was never considered to be completely solved. The earlier TF30 is simply at the lower limit of acceptable engine performance for the airframe and supported systems.

The current production model of the F110 is the -129 rated at 29 klb in reheat, with 31 klb demonstrated. The latest variant is the -129 Enhanced Fighter Engine (EFE), rated at 32 klb reheated, 34 klb demonstrated, with growth to 36 klb. The latter employs a more efficient blisk (ie single part) fan, and a simplified afterburner with a 25% lower part count. GE claim that derating the -129 EFE to about 29 klb provides a 50% increase in the required service interval. Unit cost for the F110-129 is in the range of USD 3.5-4M, based on published figures.

Whereas an upgrade to an arbitrary new engine would be quite demanding in terms of engineering effort, this is much less so the case with the F110, since the exercise was carried out by the US Navy on the F-14, replacing an engine almost identical to that being now fitted to the F-111. It follows therefore that with the experience gained by GE on the F-14 retrofit, and the RAAF with the P-109RA retrofit, the risks and engineering overheads of an F110 retrofit can be minimised.

Given the F-14 experience, the modification effort would lie primarily in the insertion of a tailpipe plug, and adapting the inlet and new nozzle to the airframe. The other required modification will be rescheduling the variable inlet. The F110 massflow requirements are only slightly greater than those of the TF30.

A refit to the F110 is not a new idea by any means, and was on the USAF F-111 squadron wishlist for many years. During the early nineties, GE proposed a refit to the USAF, and conducted preliminary engineering work. While exact numbers are not available, the expected engineering costs were of the order of USD 15-25M, and the estimated savings on support costs would pay for the upgrade in a 5-10 year in service period. The upgrade plan collapsed when the USAF decided on the early retirement of their F-111s. An upgrade initiated at this time would take about 3-4 years to complete.

The only other modification which may be required is reskinning the wing leading edges with a more durable material (eg stainless steel) to better accommodate sustained high speed operations.

In summary a refit of the F-111 fleet with an F110-GE-129 EFE is a major investment, but one which would produce some extremely useful improvements to the aircraft's basic aerodynamic capability, and signatures. This while reducing medium to long term support costs and demands for tanker support, the latter not a trivial saving with new build tankers worth about USD 130M per unit. A refit removes any system growth limits imposed by bleed air shortages in the TF30. The performance and signature gains in turn produce a major gain in survivability, especially when dealing with a fighter threat. Combining the use of internal stores with an F110 engine upgrade would result in truly blistering sustained speed performance at all altitudes. With a very useful combat radius improvement into the bargain, the case for a refit to the F110 is compelling.


The next major area to be explored is that of signatures. The principal issues here are the radar emissions from the TFR, the attack radar and the radar skin paints reflected off the aircraft.

The TFR is a robust and proven, incrementally upgraded sixties design. It is vital to low level penetration at night. However, it also broadcasts the aircraft's presence for tens of miles, not an issue in times past, but likely to become an issue in this day and age of competitively priced widely available EW equipment. Other than forgoing the low level profile, the only option here is to change the signal format to something less detectable. The expensive off-the-shelf solution would be to repackage the US Lantirn/APQ-174 navigation pod radar to wholly replace the existing AN/APQ-171 TFR. However, it may be also possible to modify or redesign the existing TFR receiver and transmitter to accommodate an LPI rather than straight pulsed waveform. Given the tactical payoff, it would be well worth investigating, and something well within the capabilities of DSTO. Alternately, the TF function could be subsumed by a dual redundant attack/TF radar.

The emissions from the AN/APQ-169 attack radar, or any conventional replacement, are an unavoidable evil. While LPI radar technology is on the threshold of large scale deployment, it will be initially expensive. However, a newer technology radar with lower peak power output, less detectable waveforms, and a low sidelobe planar array antenna would be a distinct improvement over the existing ARS. If such a radar is used sparingly, and only in Synthetic Aperture spot mapping mode for for weapon delivery, the angular extent of its transmissions will be be much smaller, and it will be significantly less detectable than the AN/APQ-169.

Reducing the radar signature of a conventional aircraft like the F-111 is not trivial, and will never provide the kind of stealth which comes with a true stealth airframe. However, dropping the forward sector signature down to a tenth of its existing size almost halves detection range and warning time for an opponent, so any dollars spent in this area are an excellent investment.

There are a number of straightforward measures which are widely used in other types in service. The canopies can be conductively coated, radar absorber can be packed into the forward radar bay, a planar array antenna can be used, a "tuned" radome can be fitted, opaque to out-of-band radars, absorbent coating or laminates can be applied to the inlets and leading edges, and critical other parts of the airframe, and finally, the aircraft can penetrate using only internal stores. All of these measures involve established technology, but would require significant R&D effort by DSTO or a contractor to implement.

Clearly the issues of speed performance and signatures are complementary, in that both are intended to compress an opponent's engagement timeline to the point, where the defence fails to perform effectively. Incremental improvements in both areas will therefore yield a nett dividend which is arguably much better than application of either measure alone. Moreover, some measures such as internal weapon carriage yield very useful returns in both areas. In summary, every dollar spent on performance improvement and signature reduction translates directly into a survivability improvement.

Radar Warning Capabilities

Electronic combat sensors are another critical aspect of survivability. In penetrating hostile airspace, an F-111 in 2010 will most likely encounter a wide range of threat radars, many with much lower peak power output than Cold War SovBloc systems and clones thereof, and most likely also a smattering of types with basic LPI (Low Probability of Intercept) features.

The only robust means of dealing with such a capability level is to have a channelised receiver for the mid and upper radar bands with the sensitivity and signal processing capability to accurate detect and track such threats.

With modern threats, knowing the range of the emitter is very useful, since it provides a much better indication of the threat's possible engagement envelope. Skirting about the edges of a SAM system's coverage is clearly much less dangerous than trying to cut through the heart of its coverage. If you penetrate faster, you require more warning time to react.

Other important benefits accrue from knowing the precise range and direction to a threat emitter. Anti-Radiation Missiles can be fired reactively in optimal range known modes. Moreover, the Harpoon (and AGM-142) can be targeted without prior radar emissions in Range-Bearing Launch (RBL) mode, allowing much more precise selection of targets, while denying the victim early warning of an impending attack. Most importantly, the ASRAAM can be launched against threatening fighters in forward quarter engagement geometries.

Rangefinding adjunct receivers are now becoming available as add ons to existing EW packages. Established designs such as the F-16CJ ASQ-213 HTS and F/A-18C LMC TAS employ interferometric techniques, combined with Phase Rate of Change, Differential Doppler or Time of Arrival (PRC, DD, DTOA) techniques, requiring additional interferometric antennas.

An impressive recent development are adjunct receivers using Digital Radio Frequency Memory (DRFM) techniques, such as the LMC (formerly IBM Federal Systems) PRSS (Passive Ranging SubSystem), which are designed to use the existing RWR antenna package without change. Such receivers are "piggybacked" on to the existing RWR, sharing its antennas, and under its control via the Mil-Std-1553B mux bus. The RWR locks the DRFM receiver on to an emitter of interest, and continues to search while the position of the threat is measured. The position of moving and static surface and airborne emitters can be found.

In 1995 US tests of the PRSS demonstrator achieved accuracies of about 250 ft CEP at 40 NMI, and sufficient accuracy for range known HARM shots in under 5 seconds. The accuracy of DRFM based adjunct receivers improves significantly with high accuracy INS/GPS navigation packages, such as installed in the AUP upgrade. The PRSS package fits into a single Milspec VME package (militarised COTS, rated to level 5 tactical Mil-Std-810E), and is cited at 55 lb (cca 25 kg).

Fitting the ALR-2002 with a channelised Hi-Mid band receiver is an incremental, but non-trivial upgrade. A suitable DRFM based range/direction finding receiver can be incorporated as an adjunct. The FEB door mid-hi band antenna layout is particularly well suited to an adjunct receiver package, providing for bearing and elevation coverage. If proper provisions are made in the implementation of the Echidna suite, incorporation of these two capabilities can be performed as intermediate upgrades to the basic AIR 5391 package.

In summary an LPI threat detection capability in the ALR-2002 will be essential, while an adjunct precision direction finding and rangefinding package would significantly improve survivability by facilitating early evasion of threats, improving the lethality of defensive anti-radiation weapons, providing a means of countering emitting hostile fighters in forward quarter geometries, and allowing silent attacks on emitting targets such as shipping.

Part 3 of this series will continue our discussion of the long term upgrade options for the F-111. Special thanks to Kurt Todoroff (formerly Capt, USAF; F-111D / EF-111A pilot, instructor, and flight examiner) for his helpful comments on the draft of this paper.


Pic.1 survivability.eps

Pic.2 F-111-Int-Bay.eps

Pic.3 engines.eps

The USN fitted a portion of their F-14A fleet, and all new build F-14D with the current technology GE F110-GE-400 reheated turbofan, replacing the sixties technology P&W TF30 fans. While the upgrade required minimal modifications to the airframe, the results were outstanding - producing a 43% increase in reheated thrust, a 37% increase in dry thrust, 30% lower fuel burn in reheat, 62% greater intercept radius, and about a 30% improvement in endurance on station and combat radius, all with a much more reliable engine with unrestricted handling. Variants of the F110 are now widely used on the F-16C, are the standard powerplant on the F-14B/D, and have been tested on the F-15E.

Pic.4 GE F110 Engine (not enclosed)

The GE F110-GE-129 IPE is the principal powerplant used in the late model F-16C. Originally designed as a powerplant for the B-1A bomber, the F110 was revised and became a competitive engine to the P&W F100 series. With a modern digital engine controller, the engine has superb handling and a high tolerance for poor inlet airflow. The latest developments of this engine deliver static afterburning thrust levels in excess of 32,000 lb, with growth to 36,000 lb, with SFC performance superior to the TF30 in all regimes (GE).

Pic.5 JDAM (not enclosed)

The GBU-31 JDAM will shortly become the USAF's standard guided bomb, carried by the B-2A, B-52G/H, B-1B, F-15E and F-16C. The baseline weapon provides precision or near precision accuracy in all weather conditions, and is a fully autonomous launch and leave weapon, with a delivery and carriage envelope virtually identical to the Mk.83/Mk.84 Slicks. The tailkit is available for the 1,000 lb Mk.83 and BLU-110, and the 2,000 lb Mk.84 and BLU-109. Depicted is a USAF B-1B rotary launcher carrying the bunker busting BLU-109/B variant of the JDAM. The JDAM would be the ideal candidate for an internally carried all weather guided bomb for the F-111C/G (USAF).

Pic.6 F-111 (not enclosed)

Retrofitting the F-111C/G fleet with a variant of the GE F110-GE-129 is a low risk proposition, given the prior TF30 retrofit effort to the F-14. This powerplant would be not only cheaper and easier to maintain in the longer term, but would also provide important gains in sustained high speed performance and combat radius. A conservative estimate based on the F-14 experience suggests that an F-111 carrying a pair of internal bombs could achieve a combat radius approaching 1400 NMI, avoiding the need for expensive tanker support in many situations. LOGO


bulletin board
| images | sounds | movies | other links | published articles | history | troops | comics | models
| external differences | patches and badges | books | AGM-142 Popeye | museums | memorabilia
Tail Numbers:
| EF / F-111A | F-111B | RF / F-111C | F-111D | F-111E | F-111F | FB-111A, F-111G | F-111K
F-111 Down-Under | airshows | RAAF Tail Numbers | ejection | artwork | feedback | Notice Board | MEMORIAL

Also visit' s companion site,   FB-111A Switchblade

Copyright '   F-111 Aardvark'.

Links to, and reviews of this site are welcome.

Disclaimer: This F-111 Aardvark Internet site does not represent the views of General Dynamics, Lockheed Martin, the United States Air Force, the Royal Australian Air Force or any other company or organisation which may be named herein.  Should any company, organisation or individual feel grieved that I am using their logo or product without permission, please contact me at the email above.

Flag Gifs from